Holocene Landscape Evolution of an Estuarine Wetland in Relation to Its Human Occupation and Exploitation: Waasland Scheldt Polders, Northern Belgium

Netherlands Journal of Geosciences — Geologie en Mijnbouw |96 – 1 | 35–62 | 2017 doi:10.1017/njg.2016.24

Holocene landscape evolution of an estuarine wetland in relation to its human occupation and exploitation: Waasland Scheldt polders, northern Belgium

T. Missiaen1,∗, I. Jongepier2,K.Heirman1,5,T.Soens2, V. Gelorini3, J. Verniers3, J. Verhegge4 & Ph. Crombe´ 4

1 Renard Centre of Marine Geology, Ghent University, Krijgslaan 281 S8, B9000 Ghent, Belgium 2 Department of History, University of Antwerp, Stadscampus, S.R-A.112, Rodestraat 14, B2000 Antwerp, Belgium 3 Palaeontology Research Unit, Ghent University, Krijgslaan 281 S8, B9000 Ghent, Belgium 4 Department of Archaeology, Ghent University, Sint-Pietersnieuwsstraat 35, B9000 Ghent, Belgium 5 Currently at Geological Survey of Denmark and Greenland, Ø. Voldgade 10, DK-1350 Copenhagen, Denmark ∗ Corresponding author. Email: [email protected]

Manuscript received: 7 October 2015, accepted: 13 June 2016

Abstract

This paper describes the landscape evolution of the Waasland Scheldt polders in the north of Belgium from the Late Glacial – early Holocene to the present time, and the effects of this changing landscape on the human settlement. The regional landscape evolution has been visualised in a series of palaeogeographical maps for successive time frames. Two different map series were produced: a series of Holocene palaeogeographical reconstructions (11,000–950 cal BP) based on geotechnical, geological and archaeological data, and a series of post-Medieval landscape reconstructions (16th- to 19th-century) based on historical maps, land registers and soil data. Additional palaeoenvironmental information from fossil pollen and plant remains allowed reconstruction of the vegetation and wetland changes, particularly for the middle to late Holocene. Peat growth was the main key to understanding the landscape evolution of the Waasland Scheldt polders. Whereas the landscape evolution during the Holocene was mainly sea-level driven, the transformation of the landscape during the last millennium was largely due to human interventions.

Keywords: historical maps, palaeogeography, peat growth, Scheldt estuary

Introduction intensively exploited (Rippon, 2000). In Romney Marsh in SE England, one of the largest coastal wetlands in Britain, research The significance of coastal and estuarine areas for understand- has allowed reconstruction of the landscape evolution and hu- ing former human life and palaeolandscapes is now recognised man exploitation from later prehistory to the medieval period internationally. For example, in the context of present-day cli- (Rippon, 2002). In the Netherlands many studies have been mate warming and sea-level rise, the study of the response of carried out in coastal and estuarine/fluvial wetlands, ranging coastal and estuarine palaeolandscapes to postglacial sea-level from Zeeland in the southwest to the Wadden Sea area in the rise is particularly relevant (e.g. Boski et al., 2002; Woodruff north, unravelling the geographical, morphological and envi- et al., 2013). The large preservation potential of these sedi- ronmental changes of these landscapes through time and the mentary environments, on the transition of the terrestrial and impact on human occupation (e.g. Van der Spek & Beets, 1992; marine environment, makes them ideal for studying landscape Vos & de Wolf, 1993; Vos & van Heeringen, 1997; Bos et al., evolution through time. Research into the intertidal area of 2005; Hijma & Cohen, 2011;Vos&Knol,2015; Vos et al. 2015). the Severn Estuary, SW England (Bell, 2007), for instance, has In Flanders, systematic Quaternary geological, sedimentolog- provided the first human Mesolithic footprints, while in Roman ical and palaeoecological research on fluvial and coastal wet- and medieval times these dynamic estuarine landscapes were lands has been carried out for a number of decades (e.g. De

C Netherlands Journal of Geosciences Foundation 2016 35

Downloaded from https://www.cambridge.org/core. IP address: 170.106.33.14, on 26 Sep 2021 at 02:43:56, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/njg.2016.24 Netherlands Journal of Geosciences — Geologie en Mijnbouw

Muynck, 1976; Augustyn, 1977, 1985; Baeteman & Verbruggen, influenced by rising groundwater due to sea-level rise, which 1979; Heyse & De Moor, 1979; Baeteman 1991, 1999;Denys, turned the area into a continuously expanding peat marsh. A 1993), and many geomorphological, geological and soil maps second reason was the strong intertwining of landscape and have been made of Belgium including its wetlands (e.g. Jacobs human occupation during medieval and post-medieval times, et al., 1993, 2010; De Moor & van de Velde, 1995; Bogemans especially in view of the great inundations of the 14th–16th 1997;AGIV,2000; Adams et al., 2002). Moreover, early recon- centuries. Both direct and indirect human interventions greatly structions of the historical landscape of the Scheldt polders influenced the (often very rapid) transformation of the land- started in the 1960s (e.g. Snacken, 1964; Mijs, 1973;Guns, scape. Lastly, the Scheldt polders are under imminent threat 1975), and research on Late Pleistocene and Holocene deposits from commercial activities. Due to the continuous expansion has been carried out here since the late 1980s (e.g. Meire & of Antwerp harbour, only a relatively small part of the original Kuijken, 1988; Kiden, 1989; Verbruggen et al., 1996;Kiden Waasland Scheldt polders still remains. A new dock is planned & Verbruggen, 2001). However, systematic geoarchaeological in this area within the next few years, while on both sides of research into onshore wetlands in Flanders is quite a recent the border the coastal realignment in the Hedwige and Prosper development. Large-scale interdisciplinary wetland research in polder will affect the last relicts of this drowned landscape, the Scheldt floodplain was often conducted in anticipation of for example through local erosion of channels, but most of the large infrastructural works such as Antwerp harbour expansion area will be further covered and preserved under new estuarine (e.g. Minnaert & Verbruggen, 1986; Gelorini et al. 2003, 2006; deposits. Perdaen et al. 2004; Crombé, 2005; Deforce et al. 2005; Meers- The main objective of this paper is to map the palaeoland- schaert et al. 2006;Deforce,2011), nature development and scape evolution of the Waasland Scheldt polders from the Late water management projects (Bogemans et al., 2012; Meylemans Glacial – early Holocene to the present time. This is done on et al., 2013). two different timescales: (1) a Holocene timescale, resulting Drilling techniques for mapping and assessing the buried ar- in a series of palaeogeographical reconstructions mainly based chaeological and palaeoenvironmental heritage were applied in on geotechnical, geological and archaeological data; and (2) a Flanders for the first time in the mid-1990s, for example in the post-medieval timescale, resulting in a series of landscape re- Verrebroek dock in the Scheldt polders (Crombé & Meganck, constructions mainly based on historical maps, land registers 1996). Since then further testing mainly in the Scheldt flood- and data of the soil mapping. Where possible, palaeolandscape plain and polders has resulted in more refined drilling tech- reconstructions are also included, based on various environmen- niques and methods (e.g. Bats, 2007; Crombé & Verhegge, tal data (pollen analyses, plant remains, etc.). By combining 2015). Recently, a new step forward was taken in prehistoric these different techniques and methodologies we were able to landscape reconstruction for archaeological purposes with the obtain a coherent picture of the drowning of the dynamic land- PhD research by Verhegge (2015). He developed an efficient scape of the Waasland Scheldt polders since the Late Glacial, approach based on near-surface geophysical and geotechnical and the effects of this drowning on the successive stages of techniques to map the prehistoric landscape of the Scheldt human settlement and land-use through time. polders, and modelled the peat growth and the subsequent drowning of the landscape. However, his research only focused on a small test area (Doelpolder Noord), and a broader regional Study area approach was still lacking. A second new development was the reconstruction of intertidal landscape response since the 16th General background century by Jongepier et al. (2015a, b). Previously this had only been attempted on short timescales, mostly less than 100 years. The Waasland Scheldt polders consist of a flat, low-lying re- Using a combination of historical maps and analysis of present- gion on the western bank of the river Scheldt, in NW Belgium day soil texture this allowed mapping of the step-wise evolution (Fig. 1). The western and eastern limits of the study area are (location of tidal channels, tidal flats and salt marshes) over respectively formed by the Dutch/Belgian border and the river the last c. 400 years of the Waasland Scheldt polders marked Scheldt, with its southern limit situated at the edge of the by de- and re-embankment (Jongepier et al., 2015b). Waasland subcuesta. The current landscape of the Waasland The Waasland Scheldt polders were selected as the study Scheldt polders is highly influenced by the proximity of the area for three reasons. First, they are known to be rich in well- North Sea and the river Scheldt. The delicate balance between preserved prehistoric sites and landscapes, as demonstrated sea-level rise, tidal regime and river sedimentation during by recent research (e.g. Crombé, 2005). Covered by 1–4 m of the Holocene resulted in different transgressive and regressive clayey and peaty deposits lies a well-preserved palaeo cover- events. Adding to this since the Middle Ages, the impact of man sand landscape which was mainly formed near the end of the on the landscape has become dominant by the building of dikes latest Ice Age; within this palaeolandscape many prehistoric and rebuilding after sporadic inundations. This battle between camp sites have been discovered. Gradually this landscape was man and water left many traces still visible in the landscape.

cupied) land due to human-induced drainage and also (at a later stage) the creation of polders (Snacken, 1964; Mijs, 1973; Augustyn, 1977;Soens,2013). The low altitude of the land is largely the result of the drainage of the peat (in addition to peat extraction) with subsequent subsidence of the land. The lowest altitudes are often related to old creeks, that either still contain water or have dried up. In general the younger polders have a higher elevation as they silted up during a longer period of time and land subsidence started later (De Kraker, 2006; Jongepier et al., 2015b;Vos,2015). Due to the continuous expansion of Antwerp harbour the polder landscape is only in parts preserved. The construction of large docks and adjacent industrial areas locally increased the original elevation by up to 10 m (see Fig. 2).

Evolution of the river Scheldt

At the end of the last glacial (c. 30–14.5 ka cal BP) the river Scheldt formed part of a braided river system that drained through the wide Flemish valley towards the west and north (Kiden & Verbruggen, 2001). The braided rivers were marked by Fig. 1. Overview of the Waasland Scheldt polders in northern Belgium (back- wide, but shallow, mostly sandy river channels with seasonally ground map from Google EarthC ). The red and black boxes indicate respec- variable water levels (Kiden & Verbruggen, 2001; Kiden, 2006). tively the extent of the Holocene (Fig. 8) and the post-medieval maps The sparse tundra vegetation cover created a surface extremely (Fig. 12). Blue and green boxes mark the extent of Figures 3 and 10.The susceptible to wind erosion (Verbruggen et al., 1996) which led grey dashed line marks the border between Belgium and the Netherlands. to the formation of local coversand ridges (Heyse & De Moor, Full and dashed black lines respectively mark existing and former dikes. 1979), a process which continued during the cold Dryas stadials Numbers and letters refer to sites (polders and docks) discussed in the of the Late Glacial (Crombé et al., 2012). One of these coversand text. 1 = Doelpolder; 2 = Sint-Annapolder; 3 = Kallopolder; 4 = Polder ridges, the Maldegem–Stekene ridge (3–4 m high and 2–3 km van Haendorp; 5 = Konings-Kieldrechtpolder; 6 = Oud-Arenbergpolder; 7 = wide; see inset Fig. 1), gradually dammed the Flemish Valley, Nieuw-Arenbergpolder; 8 = Prosperpolder; 9 = Hedwigepolder; 10 = Polder forcing the rivers to follow a new course (Kiden, 1991; Kiden & van Namen; A = Deurganck dock; B = Vrasene dock; C = Verrebroek dock; Verbruggen, 2001; Crombé et al., 2013). (This coversand ridge D = Waasland dock. ran over a much larger distance than its name seems to suggest: from the North Sea coast (Gistel) to the Waasland Scheldt polders (Verrebroek) (see Fig. 1).) Also the river Scheldt estab- The present-day Western Scheldt forms the southern part of lished a new (eastern) route, breaching through the cuesta near the Rhine–Meuse–Scheldt region, and evolved from the Honte Antwerp, possibly using an existing depression in the cuesta tidal basin during the Middle Ages (Vos & van Heeringen, 1997; (Kiden, 1991), towards the Rhine–Meuse valley. De Brouwer et al., 2001). Just north of the Waasland Scheldt Rising temperatures during the Late Glacial (14.5–11.5 ka polders lies the only remaining extensive tidal flat in the West- cal BP) caused major hydrological changes, affecting the dis- ern Scheldt, the (Drowned) Land of Saeftinghe (Fig. 1). It con- charge, regime and sediment load of the river systems (Kiden, sists of approximately 3000 ha of salt marshes, mudflats and 1991; Verbruggen et al., 1996; Bogemans et al., 2012; Crombé sand flats, cut by numerous tidal channels and creeks (e.g. Di- et al., 2013; Meylemans et al., 2013). The braided pattern of jkema et al., 1984; Meire & Kuijken, 1988; Missiaen et al., 2008; the river Scheldt changed into a large-scale meandering pat- Wang & Temmerman, 2013). tern, incising the previously infilled Pleistocene topography The present surface elevation in the Waasland Scheldt pold- (Verbruggen et al., 1991; De Moor & van de Velde, 1995; Boge- ers varies roughly between c. 0.5 and 6 m TAW (Belgian datum mans, 1997). At the start of the Holocene (c. 11.5 ka cal BP) approximate to lowest astronomical tide (LAT) at Ostend) (see climatic warming resulted in an increasingly dense vegetation Fig. 2). This implies that the majority of this region would be cover, decreasing the river discharge and sediment transport flooded (sometimes even at low tide) in the absence of dikes. (Kiden & Verbruggen, 2001). At this time the river Scheldt still In the Early Middle Ages (AD 500–1000) this region was a peaty drained towards the north into the Rijn/Maas valley (Vos & van wetland environment that progressively changed into dry (oc- Heeringen, 1997).

Fig. 2. Distribution of data used to reconstruct the Holocene evolution of the Waasland Scheldt polders (each dot represents a sediment core, archaeological augering or CPT). Background elevation data (in m NAP and TAW) from AGIV (Agentschap voor geograﬁsche informatie Vlaanderen) C . Dark blue dots indicate data points that reach the pre-Holocene deposits. Light blue dots indicate data points that were too shallow to reach the top of the Pleistocene deposits. Green dots indicate locations of 14C samples (after Verhegge et al., 2014). The black rectangle indicates the location of the peat/clay sequence at Doel-Deurganck dock that was used for multi-proxy palaeoenvironmental analysis. The white dashed line marks the border between Belgium and the Netherlands.

Around 7400–6300 cal BP the river Scheldt established a new the Honte sea branch gradually enlarged, probably as a result northwesterly route towards the North Sea through the East- of various floods (Gottschalk, 1984). Until the 15th century, ern Scheldt (Oosterschelde) (Kiden, 2006). This change of the however, the Honte connection (now called Western Scheldt) river’s position and the further rising sea level caused the Lower remained very shallow and navigation was only possible during Scheldt to turn brackish and to experience tidal influence; it high tide (Brand, 1983). Storm surges in the 15th and 16th is the furthest marine incursion for the Lower Scheldt during centuries resulted in large-scale inundations and an increase in the middle Holocene (Kiden, 2006). From roughly 5700 cal BP, the tidal regime of the Western Scheldt. This led to a shift in sea-level rise started to slow down and the tidal influence in the watershed between the Western and Eastern Scheldt, the the Lower Scheldt disappeared until the Early Middle ages. Western Scheldt now becoming the main branch of the river At least until the Early Middle Ages the river Scheldt dis- Scheldt (van der Spek, 1994; Vos & van Heeringen, 1997;Vos, charged through the Eastern Scheldt, a peat-covered ridge 2015). northwest of the Land of Saeftinghe blocking a more western course (Van Rummelen, 1965;Vos,2015). The connection Geological setting between the Honte tidal basin (the precursor of the Western Scheldt) and the river Scheldt east of Saeftinghe most likely In the Waasland Scheldt polders the Quaternary deposits rest came into existence in the 9th century AD (Leenders, 1986;Vos on Neogene sediments which consist largely of sandy deposits & van Heeringen, 1997). During the 11th and 12th centuries (Formations of Lillo and of Kattendijk), covering thick clay

beds of Oligocene age (Formation of Boom (Member of Putte)) 1995; Gelorini et al., 2006;Deforce,2011; Deforce et al., 2014a). (Jacobs et al., 1993, 2010). The Quaternary deposits are less In some places this marine incursion eroded the basal peat. than 5 m thick in the southwest and increase up to a thickness The peat deposits are overlain and locally eroded by a se- of 25–30 m in the northeast. quence of late Holocene estuarine sandy and clayey sediments, The Quaternary stratigraphy of the Waasland Scheldt pold- often with remains of organic matter or marine shell fragments, ers is complex, and over the years several subdivisions have that was deposited in a tidal flat environment (Kiden & Ver- been described and proposed for different areas (see De Moor & bruggen, 2001; Kiden, 2006). Consequently there is a lot of van de Velde, 1995). The Quaternary deposits in the Waasland lateral variability within this deposit ranging from a thickness Scheldt polders were all deposited in a dynamic environment, of roughly 5 cm up to 10 m. The most recent sediments con- implying much lateral variation within the same depositional sist of late and post-medieval flood deposits made up of (often unit. Consequently, these deposits have been catalogued into organic-rich) clay, which are locally more sandy towards the units based on the following criteria adapted from De Moor base. (2002): (1) the lithostratigraphy (including lateral extent), (2) the chronostratigraphy, (3) the lithology and sedimentology Occupational history of the sedimentary facies, and (4) the genesis of the deposit and indications for its palaeoenvironment (see Methodology The Waasland Scheldt polders are known to be rich in archae- section). ological remains, especially dating back to prehistoric and me- The oldest Quaternary deposits in the study area are Middle dieval times (Crombé, 2005; Meersschaert et al., 2006). During Weichselian sandy river deposits. They are only observed in the last decades various archaeological salvage excavations in the eastern and extreme western part of the study area. They the vicinity of Doel and Verrebroek (for location see Fig. 1), consist of fine to coarse river sands that were deposited by conducted in the context of harbour expansion, have revealed a a braided river system in a periglacial environment c. 30,000 number of well-preserved prehistoric settlements, all located on years ago De Moor & van de Velde, 1995; Bogemans, 1997; the tops and flanks of Upper Pleistocene sand ridges (Crombé, Adams et al., 2002). In the central part of the study area Late 2005). The oldest remains date back to the Final Palaeolithic Glacial and Holocene deposits lie directly on top of the Neogene and Early Mesolithic, when the landscape was still a largely dry formations. environment (Crombé et al., 2011, 2013). A series of sites dat- The covering Quaternary unit consists of Late Glacial aeolian ing back to the Mesolithic–Neolithic transition (Crombé, 2005; sand deposits (marine istope stage 2, c. 30–15 ka cal BP). Dur- Sergant et al., 2006), and attributed to the Swifterbant culture ing this period the climate was still very cold and windy, the (Crombé et al., 2011), are contemporaneous with a period of vegetation cover was limited and a thin layer of sand (on aver- increased tidal influence (Verhegge et al., 2014). age 2 m thick) was deposited over the entire Waasland Scheldt So far no direct archaeological proof of human activity has area (De Moor & van de Velde, 1995; Bogemans, 1997;Adams been found that dates from the Middle Neolithic to the Mid- et al., 2002), similar to many other regions in NW and central dle Ages, when the area was covered by large fens and peat Europe (e.g. Kasse, 2002). bogs, but archaeological records from nearby locations in the The Pleistocene coversand deposits in the Waasland Scheldt southwestern Netherlands indicate that occupation took place polders are locally overlain by Late Glacial / early Holocene even in these wet situations (De Clercq, 2009; De Clercq & Van meandering river deposits consisting of one, sometimes two, Dierendonck, 2009). For instance, near Borsele a Roman set- fining-upward cycles (from fine sand to silt/clay) and ranging tlement was discovered on top of the peat (Sier, 2003). At in thickness between 2 and 5 m (Bogemans, 1997). In other, less Colijnsplaat in the Oosterschelde estuary and at Serooskerke, energetic parts of the floodplain, clay was sometimes deposited Roman occupation was attested at the top of the peat (De (De Moor & van de Velde, 1995). Clercq & Van Dierendonck, 2009; Dijkstra & Zuidhoff, 2011). The lowermost Holocene deposits consist of (dark) brown According to Vos & van Heeringen (1997) the oldest occupation peat. Most of the basal peat accumulated in a marsh envi- of the peat landscape occurred along the edges of the estuarine ronment along the Scheldt river and estuary. With time, peat system and can be dated back to the early Iron Age (roughly also started to grow in higher locations. The total thickness 2600 BP). of the peat deposits ranges roughly between 0.1 and 6 m. In The medieval occupational history of the Waasland Scheldt the (north)eastern part of the Waasland Scheldt polders, the polders has not been completely established so far. Historical basal peat is covered by a grey to almost black clay (occa- sources inform us of a gradual intensification of land use in sionally sandier), which often contains peat fragments. This the 12th century, starting from the Waasland subcuesta in the sediment was deposited during the marine incursion of the south and the Pleistocene sand ridges, on which the medieval middle Holocene (around 6000 cal BP) which changed the low- villages Kallo, Verrebroek and Kieldrecht (see Fig. 1)aremen- lying western bank of the river Scheldt into an estuarine tidal tioned from the 12th century onwards (Augustyn, 1977;Van landscape (Minnaert & Verbruggen, 1986; Verbruggen & Denys, Gerven, 1977). Saeftinghe in the north became a stronghold

of the count of Flanders in the 13th century, controlling nav- were considered. The data were interpreted following the crite- igation on the river Scheldt (Gottschalk, 1984). The count of ria mentioned earlier (‘Geological setting’ section), and consid- Flanders and the lord of Beveren also granted large stretches ering the most current geological knowledge of the area. Using of marshlands to abbeys, which turned them into agricultural the sedimentological description (e.g. clay, sand or peat), the estates. palaeoevironmental indicators (e.g. shells or plant remains), At the height of the medieval occupation phase in the 14th the lithostratigraphic position, the chronostratigraphic infor- century, several new settlements came into existence, many of mation (if available) and the correlation with all sites in the them related to peat exploitation and transport (e.g. Namen, close vicinity, all deposits were logged in different units (not Casuwele, Sint-Laureins) (Gottschalk, 1984). The decline of peat necessarily present at all sites). These units consist from bottom exploitation and increasing flood problems (a direct result of to top of (1) a mostly sorted sandy deposit with no biological re- the lowering of the landscape due to peat compaction and ex- mains interpreted as a Late-Glacial coversand, (2) a clay and/or traction) locally intensified the general demographic and eco- sandy clay (often absent) interpreted as an early Holocene me- nomic decline of the 14th century. At the end of the 16th cen- andering river and/or river flood deposit, (3) a (dark) brown tury large parts of the Waasland Scheldt polders were flooded as peat sometimes intercalated by a clay layer which is interpreted a result of large-scale inundations, mostly intentionally caused as a peat deposit interrupted by a marine incursion, (4) a seas part of a military strategy during the Eighty Years’ War. Only quence of sand and clay deposits with shell remains interpreted the more elevated areas (e.g. the village centre of Kieldrecht as Late Holocene estuarine deposits, (5) an anthropogenic clay and the polders of Namen, Doel and Sint-Anna) were spared the interpreted as late and post-Medieval flood deposits, and (6) inundations. In the following centuries the area lost to the sea a soil cover and/or construction deposits related to harbour was gradually re-embanked and reoccupied (Jongepier et al., activities. 2015b). The majority of the CPT data involved mechanical CPT measurements. Resolution of the mechanical and electrical CPT data was respectively 10–20 cm and 2–5 cm. It is known that the ac- Methodology curacy of mechanical CPT data can sometimes be inadequate for a quantitative analysis (Lunne et al., 1997), and interpre- Holocene palaeogeographical maps tation was therefore done with great care, and where possible (11,000–950 cal BP) also comparing with sediment cores taken in close proximity. Electrical CPT logs are generally better suited to determine the The topographical and palaeogeographical maps of the Holocene different stratigraphical layers but also here the interpreta- were created using geological and geotechnical information tion was done manually and involved a good local geologi- from a wide variety of sources: sediment cores, cone penetrom- cal knowledge. Automated classification of soil stratigraphy eter tests (CPT) and archaeological augerings. The vast majority was abandoned since this often did not allow the peat and of the core and CPT data were obtained from the subsurface (organic-rich) clay layers to be distinguished (Missiaen et al., database of the Flemish Government (Databank Ondergrond 2015). Vlaanderen – DOV). New CPT and core data were obtained in In total the dataset consisted of 6423 data inputs of which 2011–14 in Doelpolder (Verhegge et al., 2014; Missiaen et al., 5783 reached the Pleistocene/Holocene boundary (Fig. 2). How- 2015) and near Kieldrecht and Verrebroek (for location see ever, not all these data could be used to deduce a detailed Fig. 1). The depth of the CPT data varied roughly between 8 Holocene stratigraphy. Notwithstanding the careful data inter- and 30 m. Geological information from archaeological auger- pretation, most of the mechanical logs only allowed the tran- ings was provided by the Department of Archaeology of Ghent sition from soft Holocene sediment to more compact sandy University; these data were obtained in the framework of vari- Pleistocene sediment to be deduced. With regard to core data, ous projects. Average depth of the augerings ranged from 2 to only cores with a detailed sediment description could be used. 7 m whereas core interdistance generally ranged between 3 and As can be seen in Figure 2, the data coverage is not evenly 50 m (with a few exceptions up to 70–80 m). distributed. In areas of archaeological interest or with a lot of A major difficulty in the dataset was the diversity of the construction works (e.g. docks of Antwerp harbour) the data type of data (electrical and mechanical CPT, mechanical core, density is very high. In the agricultural parts of the study hand augering), the diversity in depth resolution (ranging from area data are scarcer, and the data description is often less 1 cm to over 50 cm), and the diversity of observers (geologists, detailed. engineers, archaeologists). Consequently not only the quality The distinction between late Holocene estuarine (clay) de- of the data varied greatly, but also the determination of the posits and overlying late to post-medieval flood deposits was exact depth and thickness of each sediment unit. not always straightforward. For the cores the main criterion was In the final dataset, only sites with raw data available (e.g. the presence of organic material and human traces (in which detailed sediment descriptions or original CPT measurements) case we speak of flood deposits, labelled as ‘polder clay’ by

DOV). For the CPTs the main criterion was the friction ratio (Rf) the case for large-scale maps, but even medium- and small- which was often noticeably lower in the estuarine sediments. scale maps showed rather weak correlations between date and The distinction between early Holocene river clay deposits and positional accuracy (Jongepier et al., 2015a). overlying Mid-Holocene marine clay (in the absence of a lower Based on the dates of (re-)embankments and inundations peat layer) was based on colour and the presence of marine in the region, ﬁve time slices were chosen in order to conduct shell fragments. a landscape analysis of the Waasland Scheldt polders (respec- To allow optimal integration with data from the Netherlands, tively AD 1570, 1620, 1690, 1790 and 1850) (Jongepier et al., and an easier comparison with relative mean sea-level curve 2015b). For each time slice, several maps were georeferenced reconstructions, all depths were converted from the Belgian and digitised in GIS. Choosing the most appropriate map(s) for reference level (TAW) to the Dutch reference level (NAP). In each reconstruction was largely based on the positional error, practice this meant subtracting 2.33 m from every depth or since a small error would, at least in theory, provide the most elevation. accurate depiction of that area (Fig. 3). In addition to this quantitative approach, the qualitative interpretation also played an Post-medieval landscape maps (AD 1570–1850) important role, such as topographical detail and date (Jongepier et al. 2015a). In view of the importance of the Saeftinghe area For the landscape reconstructions of the post-medieval period, with respect to the post-medieval landscape evolution of the historical maps were the main source of information. The ex- Waasland Scheldt polders, this area was included for the recon- cessively large map production in the Waasland Scheldt polders structions. (a direct result of land-surveying practices related to large em- Appropriate maps for the late medieval period are not abun- bankment works) makes historical maps a source for landscape dant (the older the map, the smaller the chance of conserva- evolution studies that can hardly be overlooked. The analysed tion). Furthermore, detailed local and regional maps have only maps were selected from a database of around 300 historical been produced in large quantities from the 17th century on- maps (16th- to 19th-century) found in the (State) Archives of wards. The map of 1575 made by land surveyor F. Horenbault Brussels, Ghent, Beveren and Middelburg. (Rijksarchief Gent, Kaarten & Plans, no. 2454), showing the Although they provide a rich source of information, histor- impact of late medieval small-scale inundations, proved to be ical maps have some serious limitations. Quality and accuracy the most suitable for the 1570 reconstruction. Mean positional may vary widely between different maps, leading to misinter- error of this map was 722 m. pretations in the palaeolandscape reconstruction when using For the 1625 reconstruction the ‘map by Coeck’ (Atlas van these maps without regarding these limitations. A vital com- Loon, Scheepvaartmuseum Amsterdam) proved very valuable, ponent of the quality and usefulness of historical maps is the showing the inundations of the late 16th century and the ﬁrst planimetric accuracy, or how well distances and locations on re-embankments in the south of the Waasland Scheldt polders these maps correspond to the actual distances and locations of in great detail. Though the geometric accuracy of this map is corresponding (present-day) features. Knowing this accuracy, limited (1383 m), elaborate georeferencing resulted in a useful it is possible to evaluate the likelihood that a reconstruction depiction of the salt marsh. Moreover it makes a clear division will accurately display the former area. between the higher and lower salt marsh. The map probably Recently a methodology was developed by Jongepier et al. dates to around 1625. (2015a) that allows one to calculate, analyse and visualise By the late 17th century an increasing number of highly the planimetric accuracy of historical maps. For the present detailed large-scale maps were made. For the 1690 reconstruc- study we have applied this methodology to evaluate the plani- tion, two high-quality local maps (Atlas of Hattinga, Zeeuws metric accuracy of 30 historical maps covering the Waasland Archive Middelburg) proved very valuable for the southwest- Scheldt polder area. As one might expect, supraregional (small- ern and eastern part of the study area. Mean positional errors scale, i.e. >1000 km2) maps generally showed the lowest ac- were mostly outstanding (as low as 53 m), though correct as- curacy, with mean positional errors (MPEs) between 500 and sessment of the MPE in certain embankment areas was not 1600 m, making their use for palaeolandscape reconstruction always possible due to lack of correspondance with the actual very restricted. Regional (medium-scale, i.e. 100–1000 km2) landscape (since the entire embankment was drowned later). maps proved to be far more accurate, often becoming more The reconstruction in the remaining parts of the study area accurate over time, although a large variation was noticeable was based on two supraregional maps with a MPE of 1006 and here. Local (large-scale, i.e. <100 km2) maps provided the high- 1507 m. est planimetric accuracy, with MPEs of <50 m, and also provided For the reconstruction of 1790 a large number of highqual- enough topographical details (especially when the maps were ity maps were available. A good example is the local map by related to embankment activities). Surprisingly, however, the land surveyor J. Coppens which shows the eastern salt marsh quality of older maps (16th- or 17th-century) could be as high near the Doelpolder and on which perpendicular distances from as or even higher than more recent maps. This was certainly the dikes to the border of the higher salt marsh were also

Fig. 3. Two maps depicting the same part of a salt marsh near Doel (for location see Fig. 1). Left a map from 1813 (ARA, Arenberg, no. 842) with a mean positional error (MPE) of 104 m, right a map from 1816 (ARA, Kaarten & plans, no. 8554) with a MPE of 33 m. The map on the right with the lowest MPE is to be preferred, on condition that the date of the map corresponds with the chosen time period.

indicated. This resulted in an excellent mean positional error of In the framework of this study, a 1 m thick peat/clay se- 103 m. The reconstruction of 1850 was based on several large- quence from Doel-Deurganck dock (for location see Fig. 2)was scale maps, produced in large series. For the Dutch parts of selected for multi-proxy, palaeoenvironmental analysis. The se- the reconstruction, the ‘Bonnebladen’ of Sealand were avail- lection of this site was based on the fact that it represents one able. For the Belgian parts, the ﬁrst cadastral surveys (Primi- of the rare peat beds located relatively far inland that contain tive Cadastral maps/Maps of P.C. Popp) proved to be the most marine transgressive deposits – all the other studied peat se- useful. quences in the Waasland polders being very close to the Scheldt river. The sequence from Doel-Deurganck dock therefore allows Palaeoecological data more insight to be gained into the vegetation shift related to the marine transgression further away from the estuary. In or- The value of traditional landscape-related biological proxies der to reconstruct local vegetation and hydrological changes (e.g. fossil pollen, phytoliths, charcoal and plant macrofossils) during the middle Holocene, palynomorphs, diatoms and sed- has been demonstrated in the palaeoenvironmental study of imentological properties were analysed. Loss on ignition (LOI) natural sediment archives and archaeological features world- (Bengtsson & Enell, 1986) was applied at 3 cm intervals across wide (e.g. Nelle et al., 2010; Mayle and Iriarte, 2014; Mercuri the sediment units to estimate the amount of minerogenic and et al., 2014; Mauri et al., 2015). These proxies have been proven organic sediment input. to be powerful tools that help elucidate past environmental and A total of 17 sediment samples at varying 3 cm (peat, in situ) climatic conditions and human responses to changing ecosys- and 10 cm (organic clay, allochthonous) intervals were pre- tem services (Birks & Birks, 2006; Nelle et al., 2010; Birks et al., pared following standard pollen-analytical procedures (Moore 2014). To ensure an accurate palaeogeographical reconstruc- et al., 1991). In each sample, palynomorph counting contin- tion, we therefore also integrated relevant information from ued until at least c. 500 terrestrial pollen grains were encoun- landscape-related proxy data (mainly fossil pollen, charcoal and tered to ensure statistical robustness of the results. However, plant macrofossils) derived from palaeo-soils, peat deposits and in four samples palynomorphs were almost completely absent archaeological features in the study area. The contemporaneity (see Fig. 4, marked with x) and, hence, disregarded for fur- and comparability of these data enabled a reconstruction of the ther analysis and interpretation. Since diatoms are not well vegetation composition and wetland changes particularly from preserved and mostly absent in peaty deposits (e.g. Gelorini the middle to late Holocene. Contrarily, proxy data recorded et al., 2006), only samples (11 in total, 5 cm interval) from the from sediment archives dating from the Late Glacial to the clayey deposit were taken for in-depth analysis in order to pro- early Holocene were far less abundant and more subjected to vide additional insights into palaeohydrological conditions and taphonomical and/or interpretative constraints. Figure 2 shows possible tidal forcing. Terrestrial plant remains from the base an overview of the sampling locations for radiocarbon dates on of the two peaty units and top of the lower peaty unit (basal bulk peat samples and small terrestrial peat macro-remains (cf. peat) were selected for accelerator mass spectrometry (AMS) 14C Verhegge et al., 2014). dating (see Table 1).

Table 1. Details of the radiocarbon dates from the peat/clay sequence from Doel-Deurganck dock.

Uncal Standard Sample Stratigraphic Elevation Cal BP 1 Cal BP 2 Lab no. BP deviation δ¹³C(‰) composition position Lat. Long. (m NAP) sigma sigma RICH-20092 6269 37 −26.6 charcoal basis of basal 51°1710.7 4°1501.5 −3.95 7250–7170 7280–7020 peat layer RICH-20091 5477 42 −23.9 Urtica dioica top of basal 51°1710.7 4°1501.5 −3.67 6310–6215 6400–6190 peat layer RICH-20093 4856 36 −26.8 Urtica dioica; basis of upper 51°1710.7 4°1501.5 −3.12 5650–5580 5660–5480 wood peat layer

Calibrated according to Reimer et al. (2013).

Fig. 4. Pollen percentage and loss on ignition (LOI) diagram from Doel-Deurganck dock (for location see Fig. 1). Shaded graphs present 10× exaggeration of original percentages.

Palaeogeographical base map and Holocene The thus created Pleistocene–Holocene boundary map re- time frame flects the original palaeorelief only when the (basal) peat is still present in the subsurface. When the basal peat has been Correct reconstruction of the Holocene palaeogeography re- eroded, assumptions about the Pleistocene surface have to be quires a reliable model of the Pleistocene surface relief. An made using a good geological knowledge of the area and of isohypse map of the boundary surface was constructed using the depositional environments. In the Waasland Scheldt pold- both geostatistical software and geological interpretation. As a ers peat was present everywhere, except for two small chan- first step an empirical semi-variogram was calculated using the nels southwest of Kieldrecht (black arrows in Fig. 5)whichare 5783 data points that reached the Pleistocene–Holocene bound- linked to marine incursions due to breaching of the embank- ary. Based on the best-fit model (in our case a directional linear ments. The current relief of the Pleistocene–Holocene bound- semi-variogram with a nugget of 0.7) and using point kriging, ary surface could be considered the palaeosurface. As can be a grid for the boundary surface was then created (XY spacing seen in Figure 5 the southern and southwestern part of the 40 m, minimum eight data points per grid cell). In order to Waasland Scheldt polders, where the coversand locally almost minimise any local artefacts (e.g. oval depressions instead of reaches the surface, is marked by a higher palaeotopogra- valleys, or a higher relief than the current relief) a combination phy (above 0 m NAP or 2.33 m TAW), while the northeastern of the gridded surface, the original data points, the digital el- part is lower (below 0 m NAP) and here the Holocene cover evation model (DEM) and general geological knowledge of the is much thicker. This topography fitted very well with the area were used to draw the final Pleistocene–Holocene boundary Pleistocene/Holocene surface of the Netherlands, though the relief map by hand using ArcGIS (Fig. 5). latter generally showed less detail due to the different scale

Fig. 5. Final relief map of the top of the Pleistocene deposits (i.e. Pleistocene–Holocene boundary) based on point data, gridded data, digital elevation model and general geological knowledge. Elevation in m NAP (Dutch reference level). The red line marks a possible valley system that shows a strong link with prehistoric occupation (cf. Fig. 8A). Thick grey lines mark the location of cross-sections A to C shown in Figure 7. The black arrows mark two small channels SW of Kieldrecht where the basal peat has been eroded.

of the study (after Vos & van Heeringen, 1997; Vos et al., Pleistocene surface relief, the maps do not always follow the 2002). model of Verhegge et al. (2014) to the letter. Using the Pleistocene surface relief, together with the In order to visualise the extent and variability of the Holocene stratigraphy, different palaeoenvironmental maps for Holocene deposits three cross-sections were made that cover successive time slices could be created. The elevation of the various parts of the study area (for their location see Fig. 5). Pleistocene surface was used to determine the maximum ex- The cross-sections were created in areas with sufficient den- tent of the (Holocene) marine deposits and peat deposits. In sity of high-quality core or CPT data to allow good correlation order to obtain a time frame for the reconstructions, relative without much interpolation. In the first cross-section (Fig. 7A) sea-level curves for Belgium and the S(W) Netherlands (Denys parallel to the river Scheldt the earliest Holocene deposits (me- & Baeteman, 1995; Kiden, 1995, 2006) and a dated peat growth andering river deposits) are present in a small depression cut evolution model for the Waasland Scheldt polders (Verhegge into the surface of the top of the Pleistocene deposits. The et al., 2014) were used as they provide an age for the alti- young (1000 years old and younger) estuarine deposits are ver- tude to which the marine influence was present or show how tically and horizontally variable, locally eroding the underlying the peat expanded (Fig. 6). For the peat growth model, a se- peat. In the second, short cross-section (Fig. 7B) through the ries of radiocarbon dates from organic remains (i.e. seeds/fruits northern part of Doelpolder close to the river Scheldt the thick and charcoal) was collected at the base of the peat deposits at layer of marine deposits in between the peat deposit stands out different heights (for the sample locations see Fig. 2). Consid- clearly. In the third, long cross-section (Fig. 7C) perpendicular ering the error margins in the semi-variogram calculation (error to the river Scheldt and crossing the harbour area we see the variance of 2.5 m) and the point kriging of the model of the different sedimentary environments in the Holocene deposits

Fig. 6. Left: Holocene relative sea-level curves for the Belgian coast and the S(W) Netherlands (Denys & Baeteman, 1995;Kiden,1995, 2006). Right: Age–depth model of the base of the peat sequence in the Waasland Scheldt polders (adapted from Verhegge et al., 2014). Grey crosses indicate the age and elevation of the peat base samples collected in the Scheldt polders. The red and blue line indicate the upper and lower age envelope for this cluster of ages.

getting thinner towards the southwest, away from the river, probably pointing towards more site-specific controls (e.g. hy- where the Pleistocene coversand almost surfaces. The middle drology, basin morphometry, catchment size) on plant habitats Holocene marine clay deposits are only present in the deeper and adaptation. However, during the warmer Allerød intersta- parts. It should be noted here that not all the (channel) fea- dial (c. 13.8–12.6 ka cal BP) Pinus sp. is more present (Deforce tures of the cross-sections are equally visible on the map in et al., 2005), especially from c. 13,400–13,300 ka cal BP onwards Figure 5 since the latter is based on a generally coarser grid (late Allerød). and involved a certain amount of smoothing which may have Figure 8A shows the landscape at the start of the Holocene filtered out small topographical details. (c. 11.5 ka cal BP). Late Glacial/early Holocene channel erosion by the proto-Scheldt river and small effluents can be distinctly detected. Most likely, only the channels deeper than −4mNAP Palaeogeographical and were active river or stream channels, while the area between palaeoenvironmental evolution and −2and−4 m NAP might have been flooded occasionally during human occupation and impact heavy rainfall. In the latter area a thin (ࣘ20 cm) layer of muddy sediment, thinning out further away from the river Scheldt, can Late Glacial to early Holocene be distinguished in some of the sediment cores and geotech- (14,500–8200 cal BP) nical measurements, probably representing flood sediments. It is not unlikely that this thin mud layer may be present in Rising temperatures during the Late Glacial (c. 14.5–11.5 ka cal more places but was not detected due to the resolution of the BP) caused an increase in vegetation cover, which resulted in geotechnical data (often >10–20 cm) or to the fact that peat better soil fixation and less erosion, except for the colder Dryas growth on top obscured its presence. Moreover, these flood de- stadials (Verbruggen et al., 1996). Fossil pollen from organic posits may also have been (partly) eroded or reworked by the palaeosols, intercalating the aeolian deposits within the cover- Mid-Holocene marine incursion. This could explain why they sand region (Crombé et al., 2012), indicate that during most of do not appear on the cross-sections in Figure 7. The palaeoriver this period shallow marshy conditions locally occurred in the channels fit well with the early Holocene palaeo-Scheldt recon- study area, with Cyperaceaea (sedges) and Poaceae (grasses) as struction of Kiden (1995, 2006) as well as the early Holocene predominant herbaceous components. Surprisingly, traditional palaeoenvironmental reconstruction of the Netherlands (Vos Late-Glacial arboreal taxa, such as Salix sp. (willow), Betula sp. & van Heeringen, 1997; Vos et al., 2002)(Fig. 11A, further (birch) and Pinus sp. (pine), were less prominently observed, below).

Fig. 7. Schematic cross sections through Nieuw-Arenberg polder (A), Doelpolder Noord (B) and Antwerp harbour (C) showing the sequence of Holocene deposits overlying the coversand. A thin layer of Late Glacial / early Holocene meandering river deposits is present in some of the deeper top Pleistocene topography. The erosive power of the tidal channels and the variability of the late Holocene estuarine deposits is clearly visible. At Doelpolder Noordathick layer of estuarine clay covers the basal peat bed. For location of the sections see Figure 5.

In terms of human occupation a potential local channel run- the Kale/Durme (Crombé et al., 2011, 2013), thus underlining ning west to east (indicated by the red line in Fig. 5), following the importance of rivers as providers of drinking water and for the southern edge of the Maldegem–Stekene coversand ridge, transport during the early Holocene. seemingly had a strong attraction. This fossil river channel has been studied and sampled in a trench during archaeological First part of the middle Holocene excavations at Verrebroek ‘Aven Akkers’ (Sergant et al., 2007). (8200–7000 cal BP) Over several kilometres along both banks, but mainly along the steep northern bank, numerous sites from the Early (10,750– Rising temperatures during the early to middle Holocene re- 9350 cal BP) and Middle Mesolithic (9400–8350 cal BP) have sulted in the development of a denser forest vegetation, re- been detected during surveys and salvage excavations (Perdaen ducing soil erosion and runoff to a minimum (Verbruggen et al., 2004; Crombé, 2005; Crombé et al., 2011). Similar occu- et al., 1996). Consequently, the river discharge consistently de- pation patterns are known along other rivers from the Scheldt creased, causing gradual desiccation of the Late-Glacial flood- basin such as the Lower Scheldt (Meylemans et al., 2013)and plains. Only in the deepest channels was some shallow water

Fig. 8(A, B). Palaeogeographical maps of the Waasland Scheldt polders for different periods. (A) 11000 cal BP; (B) 7500 cal BP.

Fig. 8(C, D). Palaeogeographical maps of the Waasland Scheldt polders for different periods (continued): (c) 6500 cal BP; (D) 5000 cal BP.

Fig. 8(E, F). Palaeogeographical maps of the Waasland Scheldt polders for different periods (continued): (E) 2500 cal BP; (F) AD 1000. The map shown in (F) is highly tentative.

still present. In these areas, as part of the hydrosere process, Second part of the middle Holocene peat also started to accumulate (Kiden, 1991; Kiden & Ver- (7000–5000 cal BP) bruggen, 2001; Bos et al., 2005; Bogemans et al., 2012). During this period, the palaeo-Scheldt river flowed northward (Kiden, During the middle Holocene, relative sea-level rise dropped to 2006), while sea level was still rising rapidly, by c. 0.7 cm a−1 c. 0.4–0.25 cm a−1 (compared to 0.7 cm a−1 prior to 7500 cal BP) (Denys & Baeteman, 1995). (Denys & Baeteman, 1995; Kiden, 1995). The sedimentation The early to middle Holocene landscape of the Waasland rates were, however, relatively low due to the limited sediment Scheldt polders is illustrated in Figure 8B. The Scheldt river supply and the low transport capacity of the rivers. The tidal south of the Dutch–Belgian border was still a fresh-water en- activity in the study area was still limited, in comparison to the vironment (in contrast to the SW Netherlands where it had SW Netherlands where a shallow, lagoonal environment had al- already turned brackish by 8000 cal BP), but on the low-lying ready developed in the vicinity of the Scheldt estuary; around banks along the channels peat growth started to develop. Most 6500 cal BP the sea reached its most inland position in Zeeland likely the peat was confined to regions roughly below −4m (Vos and van Heeringen, 1997). By 6500–6000 cal BP the river NAP, around the low-lying river and stream banks. Again a strik- Scheldt had turned brackish south of the Dutch/Belgian border ingly good correlation can be observed with the corresponding (Vos & van Heeringen, 1997), and the part of the Waasland palaeogeographical map of the southern Netherlands by Vos Scheldt polders closest to the Scheldt river changed into an ex- et al. (2002) and Vos & van Heeringen (1997)(Fig. 11B, further tended tidal landscape with mudflats (including tidal channels) below). and salt marshes (Fig. 8C). Palaeoecological data from Doel-Deurganck dock indicate The limit of the marine flooding in the study area was de- that in the lowest depressions/valleys permanent wet con- termined using the occurrence of the Holocene (peri-)marine ditions occurred, favouring the development of wooded fen, deposits and the peat growth model by Verhegge et al. (2014). dominated by Alnus (alder) and Fraxinus (ash), whereas in Most of the early Holocene fens drowned and were covered the transition zone (from dry to wet) Corylus avellana (com- with an organic-rich alluvial clay (Zone II in Fig. 4). According mon hazel) occurs (see also Fig. 4, Zone I). On the higher to Deforce et al. (2014b) the lack of Phragmites (Poaceae) and – drier – elevations, i.e. the sand dunes, a predominant dry Salix may seem to suggest a fresh-water environment. How- woodland was present, with Quercus sp. (oak), Tilia sp. (lime) ever, given the presence of dinoflagellates Spiniferites and Op- and Corylus avellana (common hazel) as main arboreal com- erculodinium israelianum and an increase of Chenopodiaceae, ponents (Gelorini, unpubl. data). The pollen percentage and it seems more likely that the clayey sediments are deposited LOI diagram in Figure 4 indicate an abrupt temporary increase under brackish circumstances. This corroborates earlier stud- of Tilia around 3.8 m NAP (Zone II), concurrent with a no- ies by Minnaert & Verbruggen (1986) and Verbruggen & Denys table rise in the presence of the parasitic fungi Kretzschmaria (1995). Surprisingly, during this phase also indications of crop deusta (T. HdV-44) and Diporotheca rhizophila (T. HdV-143). cultivation (cf. cereal type) are found; however, its origin (au- This sudden change indicates the temporal development of a tochthonous/allochthonous) is unknown. Some fens, however, local dry Tilia forest phase, in which both parasitic fungi in- continued to develop, mostly confined to the transition zone fected most of the potential host populations living in the between the tidal areas and the higher Pleistocene coversands. area, and may be the main agents stimulating the Tilia dom- This peat most likely accumulated in areas below −2.5 m NAP. inance in the local forest. Hydrological stress, besides some During this middle Holocene flooding phase the transition possible damage by domestic animals, is probably the main fac- from a hunter-gatherer to an agro-pastoral economy took place. tor accelerating the disease spread. This is also suggested by a Prehistoric groups belonging to the Swifterbant Culture (c. succeeding phase, which is typified by a succession of Alnus, 6500–5950 cal BP) and Michelsberg Culture (c. 5950–5600/5500 indicating local wetter conditions (i.e. development of alder cal BP) were again attracted to the interior, settling on the same carr). coversand ridges as their early Holocene predecessors (Crombé Judging by the distribution of Late Mesolithic sites (c. 8350– &Sergant,2008). By that time these dunes were already largely 7000 cal BP), human occupation and land use changed dras- reduced in occupation surface due to peat growth and flooding, tically compared to the early Holocene. The number of sites explaining also why these small sandy outcrops are not visible decreases considerably (see Fig. 8B), indicating a reduced ex- in the landscape model. In most cases just the small top part of ploitation of the interior and/or a decrease of the group mo- the river dunes or coversand ridges was still available for set- bility (Crombé et al., 2011). Possible causes may have been the tling. This was covered by alluvial hardwood forest dominated increased density of the forest dominated by deciduous tree by Quercus sp., Tilia sp., Ulmus sp. (elm) and Fraxinus excel- species and the decreasing availability of drinking water, driv- sior (common ash) with a rich shrub layer, including Cornus ing the last hunter-gatherers to the somewhat higher and there- sanguinea (common dogwood) and Viburnum opulus (guelder fore drier banks of the Scheldt river where they settled on small rose) (Bastiaens et al., 2005; Deforce et al., 2013, 2014a; Crombé levees (Crombé et al., 2015). et al., 2015). These alluvial forests are characterised by the

highest species richness, productivity and structural and suc- ing of Ericaceae (heath), Sphagnum (peat moss) and Myrica cessional complexity within the temperate forest ecosystems gale (e.g. Gelorini et al., 2006;Deforce,2011). However, at (for references, see Deforce et al., 2013, 2014b). some sampling sites (as from, e.g., Doel-Deurganck dock and Archaeobotanical analysis (Deforce et al. 2013, 2014a) Kallo-Vrasene dock) this bog stage was preceded by a short- demonstrated that the Swifterbant groups who settled on these lived establishment of Pinus (pine) forests, probably resulting dunes mainly consumed seeds, nuts and fruits from the trees from site-specific edaphic differences (i.e. local dryer condi- and shrubs growing on the dunes, e.g. Quercus sp. (acorns), tions) (Janssens & Ferguson, 1985; Gelorini et al., 2006). Cornus sanguinea (dogwood berries), Corylus avellana (hazel- For humans, the Waasland Scheldt polders seemed much less nut), Malus sylvestris (crab apples), Prunus spinosa (sloe plums) attractive during this period, probably due to the extent of and Viburnum opulus (guelder rose berries). Thousands of cal- more open peatland, reducing the capability of settlement de- cined bone remains collected during excavations demonstrate velopment (i.e. most dunes were gradually covered by peat) that hunting and fishing were also part of the subsistence. The and decreasing the availability of food resources. Late Neolithic dominance of cyprinids among the fish remains points to the and Bronze age sites are currently only known from the dry presence of large creeks with stagnant to slow-running fresh hinterland to the west and south of the Waasland polder area water (Van Neer et al., 2013). Also the availability of Viscum al- (Thoen, 1989; De Reu et al., 2011). bum (mistletoe) and Hedera helix (ivy) may have contributed to the attractiveness of these sites. The large numbers of charcoal Late Holocene (2500–1500 cal BP) from Viscum album (mistletoe) and charred seeds from Hedera helix (ivy), collected during excavations, have been interpreted Around 2500 cal BP the coastline barriers were breached at as an indication for animal husbandry from the mid-7th millen- several locations in the SW Netherlands (Vos & van Heeringen, nium cal BP onwards (Deforce et al., 2013). Both plants are ev- 1997;Vos,2002). As a consequence, in the surroundings of ergreens, and were commonly used as winter leaf fodder during these coastal barriers peat growth ceased. The Waasland Scheldt (pre)historic times, as documented by plenty of archaeobotan- polders, however, are located much further inland, well pro- ical and historical data. tected from the invading sea, favouring the continuous accu- mulation of peat. Here, at different sampling locations, radio- Transition middle to late Holocene carbon measurements on the topmost part of the peat seem to (5000–2500 cal BP) demonstrate that peat formation probably continued – at least at some sites – until at least 1220 cal BP (c. AD 730) (Kiden, During the middle to late Holocene the relative sea-level rise 1989; Van Strydonck, 2005; Gelorini et al., 2006;Deforce,2011; decelerated from c. 0.4–c. 0.25 m a−1 to 0.07 m a−1 (Denys & Verhegge et al., 2014). Baeteman, 1995), leading to a more balanced net sedimenta- The exact extent of the peat has been a subject of debate tion. As the tidal landscape started to fill up, peatland started in Belgium between geoscientists (e.g. Verhoeve & Verbruggen, to expand seaward in a relatively short period of time. Accord- 2006) and historians (e.g. Soens & Thoen, 2009). Geoscientists ing to Vos and van Heeringen (1997), the tidal area of Zeeland expect traces of peat growth in the soil and/or subsoil. There- (southern Netherlands) was completely covered by peat in a fore, if no traces of peat are found, past peat growth is con- period of c. 500 years. In the Waasland Scheldt polders at the sidered doubtful (this viewpoint was, however, already ques- southern edge of the Zeeland region a substantial peatland area tioned by Vos & van Heeringen, 1997). Historians, on the other already existed around 5000 cal BP (Fig. 8D). This is in good hand, use information from historical records and maps and agreement with the landscape reconstruction in the southern accept more circumstantial evidence like place names or writ- Netherlands (11C, further below). ten records of peat extraction as a corroboration of peat pres- In our study area this renewed peat formation took place un- ence (Soens & Thoen, 2009). Recently, Jongepier et al. (2011) der more mesotrophic conditions (e.g. at Kallo-Vrasene Dock, showed that combining geographical and historical data can Doel-Deurganck dock). Here, the alder carr vegetation was di- help to bridge the gap between geoscientists and historians. rectly succeeded by more open sedge fens, characterised by In the eastern part of the Waasland Scheldt polders, peat Cyperaceae, Poaceae and filicales, and poor fen stages with Be- is clearly present in sediment samples and/or indicated by the tula and Myrica gale (bog myrtle) (e.g. Munaut, 1967; Janssens geotechnical data (CPT logs) (white hashed area in Fig. 9). The & Ferguson, 1985; Minnaert & Verbruggen, 1986; Gelorini et al., top of the peat layer surprisingly has a maximum elevation of 2006;Deforce,2011). Over the next 1000–1500 years the peat- c. 1 m NAP (or 3.3 m TAW) (see Fig. 7), which is exceptionally land slowly expanded further westward towards higher grounds, high for the region. Formerly, the highest point up to which with the exception of a few small ‘islands’ of coversand, which peat growth had been recorded in the Waasland Scheldt polders were nevertheless enclosed by peat. At the earliest, around ranged roughly between −1.3 and −0.8 m NAP (or 1 and 1.5 m 4000 cal BP (at Doel; see Deforce, 2011) the sedge fen was TAW) (Crombé et al., 2005; Meersschaert et al., 2006). According gradually replaced by oligotrophic peat bogs, mainly consist- to Verhoeve & Verbruggen (2006)thisisthethresholdlevel

Fig. 9. Palaeogeographical map of the peat expansion in the Waasland Scheldt polders around 2500 cal BP. This situation likely lasted till c. 1350 cal BP. The white hashed area indicates where peat was detected in the cores and/or geotechnical data. Green dots mark sediment cores with a presence of peaty sand or sand with peat fragments, but where no deﬁned peat layer was found. The black hashed area indicates where traces of medieval drainage features (‘Blockstreifen’ pattern; see also Fig. 10) are still visible.

for peat growth, as locations above 1.5 m TAW (−0.8 m NAP) aerial pictures (Fig. 10);itsextentisshowninFigure 9 (black are generally considered to be too dry. In the best case, an hashed area). Some geoscientists consider the presence of this impermeable layer in the shallow subsurface needs to be present Blockstreifen pattern insufficient to prove the existence of peat to retain rainwater in the soil. However, Ovaa et al. (1957) state (e.g. De Muynck, 1976). However, the fact that the pattern that in lower depressions between sand ridges the substrate is is visible in the medieval morphology (i.e. in areas where impermeable enough to allow peat growth and that in the past the late medieval surface is not covered by post-medieval wetter conditions must have existed in these depressions as tidal deposits) supports the historical records. The thin peat drainage was considerably worse. The depressions southeast of layer (roughly 10 cm) that was mostly left behind during the Kieldrecht seems to have fulfilled these requirements, where exploitation may easily be missed due to the low sampling traces of peat have been recorded in a number of sediment resolution in the cores (50 cm or more), or it may have cores (green dots in Fig. 8E).Sincethedatapointsforthis disappeared altogether due to drainage and oxidation, or due part of the Waasland Scheldt polders are unevenly spaced and to erosion during later inundations. It is likely that such a the sediment descriptions are based on subsurface samples at Blockstreifen pattern also existed further north, in the area a 50 cm interval, it seems likely that peat layers may not have around Kieldrecht (Augustyn, 1999), but the large floods in been detected. the 16th–17th century have wiped out all evidence. There is also a lot of historical evidence concerning peat Based on the assumptions stated above, it therefore seems exploitation. Already in the 12th and 13th centuries, peat ex- likely that the Waasland Scheldt polders landscape would have ploitation in the Waasland Scheldt polders was very significant been completely covered with peat around 2500 cal BP (Fig. 8E), (Jongepier et al., 2011). According to Augustyn (1999), around and this situation likely persisted for c. 1000 years (till c. 1500 AD 1300 the counts of Flanders realised an annual production cal BP). The map of 2500 cal BP also agrees well with the of about 8000 ‘last’ of peat – one last equalling 10,000 blocks palaeoenvironmental reconstruction of the Netherlands for this of peat – on their estates in this region. Unfortunately, the period (Vos & van Heeringen, 1997; Vos et al., 2002)(Fig. 11D, historic documents seldom mention where exactly the peat was further below). This complete peat coverage might explain dug. However, it is known that two major reclamation centres the – so far – total absence of archaeological sites belong- were founded in Kieldrecht and Verrebroek (Augustyn, 1985). ing to the Iron Age and Roman Period in the Waasland Scheldt The peat reclamations presented a so-called Blockstreifen polders. However, studies from the Belgian coastal plain (Baete- pattern with long, narrow parcels of land separated by ditches, man, 2007; Demey et al., 2013, Baeteman & Pieters, 2015)and often starting at a road or waterway (Gottschalk, 1984). This the SW Netherlands (De Clercq, 2009; De Clercq & Van Dieren- pattern can still be identified in the region on the DEM or on donck, 2009) indicate human activity in these areas that were

It seems likely, however, that the landscape during the Early Middle Ages looked similar to the landscape a thousand years earlier, except for some tidal flats and salt marshes close to the river Scheldt. Augustyn’s (1977) statement that in the Early Middle Ages the Waasland Scheldt polders consisted of a peat bog with some small sand ridges and pools in between may well be a correct description. On the present DEM (see Fig. 2) and on the soil maps some of these old creeks can still be distinguished, but it is almost impossible to determine the age of these features. Using the soil map (AGIV, 2000) and geological knowledge of the area, a tentative palaeogeographical reconstruction was made for c. AD 1000 (Fig. 8F).Bothar- chaeological and historical traces of human occupation before AD 1000 are missing. This does not imply that the area was completely uninhabited. A low-intensive land use directed at the exploitation of the wetland resources (pasturing, fishing, fowling, etc.) is possible parallel to what happened in other parts of the coastal wetlands in this period (see Soens et al., 2014). Starting in the 11th or 12th century, small-scale dams were Fig. 10. Google EarthC image of the area around Verrebroek (for location built, which either served as elevated roads in the wetland see Fig. 1) showing the presence of a medieval ‘Blockstreifen’ pattern (red area or as drainage improvements. Archaeological excavations dotted lines) in the landscape where the late medieval surface is not covered recently discovered the remains of such a dam south of the vil- by tidal deposits. Drainage of the peat lands was done by digging many lage of Kieldrecht (Cryns et al., 2014). In this period, ownership ditches, perpendicular to reclamation axes (mostly a road, indicated in over the unreclaimed ‘wastelands’ was gradually established by light blue). local lords – the lord of Beveren, whose castle Singelberg was situated immediately north of the higher cuesta (Wilssens et al., characterised by large peat bogs at that time. It is not unlikely 2007). For the 13th and 14th centuries, historical sources in- that the large-scale extraction (and erosion) of the peat may form us of the systematic reorganisation of the landscape. In have (partly) destroyed the archaeological evidence from Iron the surroundings of the villages of Kieldrecht and Verrebroek, Age and Roman sites in the study area. a pattern of dikes (so-called moerdijken or ‘peat dikes’), ditches and roads was set up in order to excavate and transport the Middle Ages (1500–500 cal BP) (AD 500–1450) extracted peat (Augustyn, 1999). Closer to the river Scheldt, marshlands were protected from As stated earlier, the man-made transformation of the land- flooding by dikes from the 13th century onwards and turned scape through dike building, draining and peat extraction into ‘polders’ (e.g. the Harnesse in Kieldrecht, protected by startedinthe11thand12thcenturies.Unfortunately,itis dikes in 1262) (Van Roeyen, 2007). In the 14th and 15th cen- still uncertain what the landscape looked like prior to these turies, larger dikes were built in order to keep the Scheldt floods man-induced landscape changes. Once drained, peat soils often out, but the water of the Scheldt increasingly invaded the low- became subject to rapid erosion and shrinkage, as documented lying region. Combined with the increasing tidal influence on for many peatland regions in the western and northern Nether- the river Scheldt, the region had become very vulnerable to lands (e.g. Borger, 1992;Vos,2015). The extraction of the peat floods as the land level in many cases was lowered through further accelerated this process. As result of this human in- the drainage, shrinking and extraction of the peat (Vos and terference and the disappearance of the top layer of peat, a van Heeringen, 1997;Soens,2013). A highly dynamic period depositional hiatus of several centuries is visible in the soil of floods, alternated with renewed land reclamation through archive from the documented end of the peat growth in the embankment set in. 7th–8th centuries AD until renewed flooding and deposition of Archaeobotanical analyses (fossil pollen and plant macro- estuarine clay deposits, locally attested in the 10th–11th cen- fossils) revealed that in the 14th–16th centuries the south- turies and more widespread in the 13th century AD (Deforce, ern part of the Haendorp polder (for location see Figs 1 and 2011). Because of both peat erosion/extraction and overlying 12) was characterised by a relatively open, agrarian land- estuarine deposits, the reconstruction of the early medieval scape, associated with crop rotation of cereals and leguminous landscape before the start of large-scale drainage and embank- crops such as peas and beans, and limited presence of wood- ment remains tentative. land. In the northern part, where peat remained still (partly)

Fig. 11. Combined palaeogeographical maps of the Waasland Scheldt polders and the neighbouring southern part of the Netherlands showing a good correlation (partly after Vos, 2002 and Vos & van Heeringen, 1997). (A) 11,000 cal BP; (B) 7500 cal BP; (C) 5000 cal BP; (D) 2500 cal BP.

uncovered, more diverse landscape types occur, consisting of among others, Hoog-Verrebroek, Kieldrecht and Doel (Van Ger- heath and grasslands, shrub and woodland vegetation (Gelorini ven, 1977). Furthermore, large parts of the peatland earlier et al., 2003). excavated were drained and converted into agricultural land. The landscape reconstruction of 1570 (Fig. 12A) shows that Late Middle Ages till modern times by then almost the entire study area was embanked and a (500–100 cal BP) (AD 1450–1850) large number of (small) villages have been founded. West of Doelpolder, remains of the former peatlands are still found. An important wave of reclamation through embankment of pre- The salt marsh is limited to the fringes of the river Scheldt vious ﬂooded land took place from 1431 onwards, when large outside of the dikes. In contrast to the earlier embankments stretches of marsh were sold by Philip the Good, duke of Bur- of the Middle Ages, which were often more curved in shape, gundy, to private drainage companies (Jongepier et al., 2012). dictated by the landscape, these later embankments became Until 1567, various embankments resulted in the polders of, increasingly linear. From the 17th century onwards the typical

Fig. 12. Post-medieval palaeolandscape maps for different time periods (adapted after Jongepier et al., 2015b). (A) AD 1570; (B) AD 1625; (C) AD 1690; (D) AD 1790; (E) AD 1850. Numbers and letters refer to sites discussed in the text. 1 = Doelpolder; 2 = Sint-Annapolder; 3 = Kallopolder; 4 = Polder van Haendorp; 5 = Konings-Kieldrechtpolder; 6 = Oud-Arenbergpolder; 7 = Nieuw-Arenbergpolder; 8 = Prosperpolder; 9 = Hedwigepolder; 10 = Polder van Namen; A = Saeftinger gat; B = Deurganck. The white dashed line marks the present-day border between Belgium and the Netherlands.

embankment layout was in regular (orthogonal) grids (Soens due to centuries of peat extraction and drainage, accompa- et al., 2014). nied by compaction, which had significantly lowered the sur- Military inundations during the Eighty Years’ War (1568– face of the land (often lower than Mean High Water Level) 1648) resulted in renewed flooding of large parts of the (Vos, 2015). This meant that large areas could easily be flooded Waasland Scheldt polders. The impact of the floods was severe once the dikes were breached. Since no immediate recovery

plans for the drowned area were made, an extensive tidal flat Holocene to the present time, and the effects of this changing developed (Land of Saeftinghe), cut by a large tidal channel landscape on the human settlement. The regional landscape (so-called Saeftingher Gat). The landscape reconstruction of evolution has been visualised in a series of palaeogeographi- 1625 (Fig. 12B) shows how this tidal flat extended far into the cal maps for successive time frames; for each map the various Waasland Scheldt polders. Only the higher areas such as the vil- driving mechanisms behind the palaeoenvironmental changes lage centre of Kieldrecht (on a sandy ridge) and the polders of and human occupation are discussed. Two different map se- Namen, Doel and Sint-Anna escaped complete flooding. Most of ries were produced: a series of Holocene palaeogeographical the villages that existed in 1570 appear to have been drowned reconstructions (11,000 cal BP – AD 1000; Fig. 8) based on now. geotechnical, geological and archaeological data, followed by In the centuries after the end of the Eighty Years’ War (in a series of post-medieval landscape reconstructions (AD 1570– 1648) the entire area was gradually re-embanked. The first areas 1850; Fig. 12) based on historical maps, land registers and soil to be re-embanked included Doelpolder and Sint-Anna polder data. The basis for the Holocene reconstructions was provided (which had largely escaped flooding) in 1614, and the polders by the top Pleistocene relief map (Fig. 5), which was used to south of Verrebroek in 1618. The landscape reconstruction of determine the maximum extent of the successive marine, peat 1690 (Fig. 12C) shows some continuity with the landscape in and estuarine deposits. A solid time frame was provided by 1625, but the course of the main tidal channel has changed relative sea-level curves and a dated peat growth evolution and it now runs due east to the Doelpolder where an internal model (for the Holocene landscapes) and old historical maps connection to the Scheldt river was established (the so-called (for the post-medieval landscapes). Palaeoecological data such Deurganck;seeFig. 12C), probably in order to facilitate future as pollen, charcoal and plant macrofossils provided information military inundations. Most of the tidal area consists of low-lying on the vegetation and wetland changes, particularly for the mudflats. Due to the successive embankments, sedimentation middle to late Holocene. The landscape of the Waasland Scheldt seaward of the new sea dikes was reinitiated after each embank- polders is highly dynamic, and only through these combined ment, leaving little time for higher salt marshes to be formed. methods was it possible to obtain an accurate reconstruction of In the course of the 18th century, embankment contin- the (drowning) landscape and to interpret successive stages of ued with the Nieuw-Arenbergpolder (in 1729–1784). Dikes be- human settlement and land use. came higher and stronger, but especially the landscape ‘design’ In short the evolution of the Waasland Scheldt polders land- changed drastically: while medieval embankments were often scape can be described as follows. At the start of the Holocene consistent with the natural topography, the early modern em- (c. 11,500 PB) the landscape was marked by coversand de- bankments were characterised by a regular pattern of perpen- posits, towards the east locally eroded by channels of the dicular roads and ditches and a rectangular parcelling, neglect- palaeo-Scheldt river. Human occupation was concentrated ing all natural features (De Kraker, 2007). Due to the larger along the southern edge of an E–W-trending sand ridge, most embanked area, the volume of the tidal area decreased, and likely the location of a former fossil river channel. Rising tem- therefore the flood and ebb discharges going through the tidal peratures during the early Holocene resulted in the gradual channel system were reduced, causing sedimentation in the development of a woodland, and peat started to grow in the channels themselves (D’Alpaos et al., 2006; Vandenbruwaene deeper channels. Human occupation decreased considerably, et al., 2012). The landscape reconstruction of 1790 (Fig. 12D) the last hunter-gatherers settling on small levees on the banks shows that, apart from the new embankments, almost a century of the Scheldt river, which was still a fresh-water environment. of sedimentation has allowed the salt marsh to be heightened As sea level rose, a large part of the area changed into an in the Land of Saeftinghe. The tidal channel is much reduced extended tidal landscape with mudflats and salt marshes dur- in size and the area of lower salt marsh has extended, but also ing the middle Holocene. Human occupation again returned to higher salt marsh has developed against the sea-dikes of most the coversand ridges, though now often concentrating on the of the embankments. top part due to extending peat growth and flooding. Already The process of salt marsh formation persisted in the period around 5000 cal BP a substantial peatland area existed, making 1790–1850. By then, almost the entire tidal flat was bordered human occupation increasingly less attractive. During the late by a salt marsh, located along the outer dikes bordering the Holocene, peat growth gradually took over the entire area. By remaining intertidal area, and the tidal channel surface reduced 2500 cal BP almost the entire area was covered by peat, which even further in size (Fig. 12E). probably explains the absence (so far) of Iron Age and Roman settlements. Peat growth probably continued till roughly 1200 cal BP. Synthesis During the Early Middle Ages the landscape was still largely peat-covered, except for some tidal flats and salt marshes close In this paper we have described the landscape development to the river Scheldt. Traces of human occupation are miss- of the Waasland Scheldt polders from the Late Glacial – early ing, but this does not exclude some local land use (pasturing,

fishing, etc.). Human intervention in the landscape started methods, which were successfully applied to reconstruct me- in the 11th–12th century with the building of small dams dieval peat colonisation in different parts of the Netherlands (for roads or drainage). From the 13th century onwards dikes, (Leenders, 1989; Borger, 1992;DeLangen,1992; Ligtendag, ditches and roads were set up to excavate and transport the 1995;deBont,2014) are problematic in regions where the peat. Closer to the river Scheldt, larger dikes were built to medieval landscape has been covered by thick layers of post- protect the increasingly invaded low-lying region. Intensive medieval sediments such as our study area. The present palaeoland reclamation through embankment took place, and large geographical reconstructions, including a detailed mapping of parts of the earlier-excavated peatland were drained and con- the Pleistocene surface relief, offer a new and solid base for verted to agriculture land. By 1570 almost the entire area was such enquiry. embanked, and a large number of villages had been founded. Peatland only occurred in the west. During the next 50 years Conclusions military inundations resulted in large-scale flooding of the area – a direct result of the increasing tidal influence and lowering The interdisciplinary reconstruction of the Holocene palaeo- of the land through drainage, shrinking and extraction of the geography and occupation history of the Waasland Scheldt peat. Many villages that existed in 1570 were drowned. In the polders presented here is quite new in Belgium. For the first following centuries the area was gradually re-embanked, and time a series of detailed palaeogeographical maps and land- the remaining tidal area pushed back to the northern limits scape reconstructions has been made that gives an overview (Land of Saeftinghe). Dikes became larger, and the embank- of both the long-term (typically thousand-year period, pre- ments were increasingly characterised by a regular pattern ne- medieval) and short-term (typically hundred-year period, post- glecting all natural features. As the tidal area decreased, the medieval) evolution of this wetland region since the Late Glacial marsh in Saeftinghe was considerably heightened. – early Holocene, including recent historical times. Previous re- In contrast to the landscape evolution during the Holocene constructions typically focused on a more limited time period which was mainly sea-level driven, the landscape transforma- (e.g. Middle to Late Holocene, or post-medieval), did not com- tion during the last millennium (i.e. since the Early Middle bine data from such a wide range of disciplines investigating Ages) was largely due to human interventions. The latter in- past landscape evolutions or did not attempt to extrapolate cluded both direct landscape modifications (such as the de- data into a coherent landscape model. velopment of a drainage and flood protection infrastructure, The maps presented in this study are based on an exten- agricultural land use or settlement) and their indirect and sive body of existing and new data. In the future this database mostly unintended consequences (such as the dramatic low- will be continually updated with new information from many ering of soil levels due to peat drainage, as well as the in- different sources (e.g. geology (boreholes), geomorphology, ar- crease of storm-flood levels in the estuary as the accommoda- chaeology, datings, palaeoecology, historical data and maps), tion space for excess flood water had shrunk due to progres- not only related to academic research but also in the frame- sive embankment (Soens, 2013;Vos,2015; and for the north- work of commercial projects (among others, the planned con- ern Netherlands, Van Dam, 2001;Knol,2013)). An important struction of a large new dock (the so-called Saeftinge dock) key to understanding the landscape evolution of the Waasland affecting large parts of Doelpolder and Nieuw-Arenbergpolder). Scheldt polders is peat and its nature, growth, coverage and This new information should allow further refinement of the extraction. On the one hand, peat has the great advantage maps, and, where necessary, their modification. Expanding the of covering and preserving former landscapes and the archae- present maps to a wider regional scale, as was done in the ological traces of prehistoric occupation it contains. On the Netherlands by Peter Vos, may seem a logical step but this will other hand, through its transient nature (due to shrinkage, require a significant effort. Nonetheless, regional palaeogeo- extraction, erosion, etc.), peat often makes landscape recon- graphical maps can be a valuable tool for the prospection of struction difficult (as traces of settlement on top of the peat buried archaeological heritage because they show which palae- have often disappeared). By combining multiple methods and olandscapes (for a specific period) favour human settlement disciplines, former interpretations of peat growth in the area, and/or specific human activities. In turn this can lead to new ar- which had been the subject of intense debate in Belgium, could chaeological data that may supply important information about now be corrected. The peat evolution in the Waasland Scheldt the palaeoenvironment and the age of deposits and which will polders correlates extremely well with the Holocene landscape help to improve the map reconstructions. maps from the SW Netherlands (Fig. 11) by Vos & van Heerin- gen (1997) and Vos et al. (2002), although the Dutch maps show less resolution due to the scale involved. More research is still Acknowledgements needed to reconstruct the chronology and topography of medieval peat reclamations and the subsequent disappearance of The presented research was funded by the EU Interreg 2 the peat. Traditional historical-geographical and archaeological Seas programme (project ‘Arch-Manche’) and the Research

Foundation Flanders (FWO) (project ‘Archaeological exploration broek and Doel Excavation Projects (Vol. 1: Palaeo-environment, chronology across the land–sea boundary in the Doelpolder Noord area: im- and features). Archaeological Reports Ghent University 3, Academia Press pact of sea-level rise on the landscape and human occupation, (Ghent): 251–78. from the prehistory to medieval times’). Iason Jongepier and Bats, M., 2007. The Flemish Wetlands: an archaeological survey of the valley Tim Soens acknowledge funding by the University of Antwerp of the river Scheldt. In: Barber, J., Clark, C., Cressey, M., Crone, A., Hale, Research Council (‘Drowned but not deserted. Interactions be- A., Henderson, J., Housley, R., Sands, R. & Sheridan, A. (eds): Archaeol- tween social and ecological resilience of estuarine landscapes ogy from the Wetlands: Recent Perspectives. Proceedings of the 11th WARP after flooding. Test-case: the Waasland polders on the west- Conference, Edinburgh 2005. Society of Antiquaries of Scotland (Edinburgh): bank of the river Scheldt (15th–18th centuries)’). Koen De- 93–100 (WARP Occasional Paper 18). force, Stijn Temmerman and Peter Vos are kindly acknowledged Bell, M., 2007. Prehistoric coastal communities: the Mesolithic in Western Britain. for their critical review of the paper. The authors would like to Council for British Archaeology Research Report nr 149 (York): 250 pp. thank the two reviewers, Patrick Kiden and Cecile Baeteman, Bengtsson, L. & Enell, M., 1986. Chemical analysis. In: Berglund, B.E. (ed.): for their helpful comments. Handbook of Holocene palaeoecology and palaeohydrology. Wiley and Sons (New York): 423–451. Birks, H.H. & Birks, H.J.B., 2006. Multi-proxy studies in palaeolimnology. Vege- References tation History and Archaeobotany 15: 235–251. Birks, H.H., Gelorini, V., Robinson, E. & Hoek, W.Z., 2014. Impacts of palaeo- Adams, R., Vermeire, S., De Moor, G., Jacobs, P., Louwye, S. & Polfliet, T., climate change 60 000–8000 years ago on humans and their environments in 2002. Toelichting bij de Quartairgeologische kaart Kaartblad (15) Antwer- Europe: integrating palaeoenvironmental and archaeological data. Quaternary pen Schaal 1:50 000. Vlaamse overheid, Departement Leefmilieu, Natuur en International 378: 3–14. Energie, Dienst Natuurlijke Rijkdommen: 50 pp. Bogemans, F., 1997. Toelichting bij de Quartairgeologische kaart Kaartblad (1–7) AGIV, 2000. Vectoriële versie van de Bodemkaart 1/20000. AGIV (Agentschap voor Essen — Kapellen Schaal 1:50 000. Vlaamse overheid, Departement Leefmilieu, geografische informatie Vlaanderen), IWT, Laboratorium voor Bodemkunde Natuur en Energie, Dienst Natuurlijke Rijkdommen: 38 pp. van de Universiteit Gent (GIS-Vlaanderen). Bogemans, F., Meylemans, E., Jacobs, J., Perdaen, Y., Storme, A., Verdurmen, Augustyn, B., 1977. Bijdrage tot het ontstaan en de vroegste geschiedenis van de I. & Deforce, K., 2012. The evolution of the sedimentary environment in the Wase Polders – van de oudste tijden tot circa 1400. Annalen van de Koninklijke lower River Scheldt valley (Belgium) during the last 13,000 a BP. Geologica Oudheidkundige Kring van het Land van Waas 80: 5–97. Belgica 15(1–2): 105–112. Augustyn, B., 1985. De turfwinnersdorpen Kieldrecht en Verrebroek in 1394: Borger, G.J., 1992. Draining-digging-dredging; the creation of a new landscape in Twee stadia in de evolutie van een proto-industriële naar een agrarische the peat areas of the Low Countries. In: Verhoeven, J.T.A. (ed.): Fens and bogs produktiewijze. Annalen van de Koninklijke Oudheidkundige Kring van het in the Netherlands: vegetation, history, nutrient dynamics and conservation. Land van Waas 88: 241–256. Kluwer (Dordrecht, Boston, Londres): 131–171 (Geobotany, 18). Augustyn, B., 1999. De veenontginning (12de-16de eeuw). Geschiedenis van volk Bos, J., Huisman, D., Kiden, P., Hoek, W. & van Geel, B., 2005. Early Holocene en land van Beveren. Gemeente Beveren, Beveren, Belgium: 172 pp. environmental change in the Kreekrak area (Zeeland, SW-Netherlands): Baeteman, C., 1991. Chronology of the coastal plain development during the a multi-proxy analysis. Palaeogeography, Palaeoclimatology, Palaeoecology Holocene in West Belgium. Quaternaire 2(3/4): 116–125. 227: 259–289. Baeteman, C., 1999. The Holocene depositional history of the palaeovalley of the Boski, T., Moura, D., Veiga-Pires, C., Camacho, S., Duarte, D., Scott, D.B. & IJzer (western Belgian coastal plain) with reference to the factors controlling Fernandes, S.G., 2002. Postglacial sea-level rise and sedimentary response in the formation of intercalated peat beds. Geologica Belgica 2: 39–72. the Guadiana Estuary, Portugal/Spain border. Sedimentary Geology 150(1–2): Baeteman, C., 2007. Roman pea-extraction pits as possible evidence for the timing 103–122. of coastal changes. An example from the Belgian coastal plain. In: Beenakker, Brand, K., 1983. Over het ontstaan en de ontwikkeling van de Hont of Wester- J.J.M., Horsten, F.H. & de Kraker, A.M.J. (eds): Landschap in ruimte en tijd. schelde. Zeeuws tijdschrift 33(3): 99–110. Aksant Academic Publishers (Amsterdam): 16–25. Crombe,´ Ph. (ed.), 2005. The last hunter-gatherer-fishermen in Sandy Flanders Baeteman, C. & Pieters, M., 2015. Hoe en waarom het landschap veranderde (NW Belgium). The Verrebroek and Doel excavation Projects. Volume 1: Palaeo- tijdens de Romeinse periode te Ra-versijde (Oostende, Belgische kustvlakte). environment, chronology and features. Archaeological Reports Ghent Univer- In: Hameeuw, H. (ed.): Recent Archeologisch onderzoek in West-Vlaanderen. sity 3, Academia Press (Ghent): 334 pp. Vereniging voor Oudheidkundig Bodemonderzoek in West-Vlaanderen vzw Crombe,´ Ph. & Meganck, M., 1996. Results of an auger survey research at the (V.O.B.o.W.) (Roeselare): 1–25. Early Mesolithic site of Verrebroek “Dok (East-Flanders, Belgium). Notae Baeteman, C. & Verbruggen, C., 1979. A new approach to the evolution of the Praehistoricae 16: 101–115. so-called surface peat in the Western Coastal Plain of Belgium. Aardkundige Crombe,´ Ph. & Sergant, J., 2008. Tracing the Neolithic in the sandy lowland of Dienst, Prof. Paper 1979/11 no. 167: 21 pp. Belgium: the evidence from Sandy Flanders. In: Fokkens, H., Coles, B., van Bastiaens, J., Deforce, K., Klinck, B., Meersschaert, L., Verbruggen, C. & Gijn, A., Kleijne, J., Ponjee, H. & Slappendel, C. (eds): Between foraging and Vrydaghs, L., 2005. Palaeobotanical analyses. In: Crombé, Ph. (ed.): The farming. An extended broad spectrum of papers presented to Leendert Louwe last hunter-gatherer-fishermen in Sandy Flanders (NW Belgium). The Verre- Kooijmans, Analecta Praehistorica Leidensia 40: 75–84.

Crombe,´ Ph. & Verhegge, J., 2015. In search of sealed Palaeolithic and Mesolithic Deforce, K., 2011. Middle and late Holocene vegetation and landscape evolution sites using core sampling: the impact of grid size, meshes and auger di- of the Scheldt estuary. A palynological study of a peat deposit from Doel ameter on the discovery probability. Journal of Archaeological Science 53: (N-Belgium). Geologica Belgica 14: 277–288. 445–458. Deforce, K., Gelorini, V., Verbruggen, C. & Vrydaghs, L., 2005. Pollen and Crombe,´ Ph., Perdaen, Y. & Sergant, J., 2005. Features. In:Crombé, Ph. (ed.): phytolith analyses. In:Crombé, Ph. (ed.): The last hunter-gatherer-fishermen The last hunter-gatherer-fishermen in Sandy Flanders (NW Belgium). The in Sandy Flanders (NW-Belgium). The Verrebroek and Doel Excavation Projects. Verrebroek and Doel excavation projects Vol. 1: Palaeo-environment, chronol- Vol. 1: Palaeo-environment, chronology and features. Archaeological Reports ogy and features. Archaeological Reports Ghent University 3, Academia Press Ghent University 3, Academia Press (Ghent): 108–126. (Ghent): 141–179. Deforce, K., Bastiaens, J., Ervynck, A., Lentacker, A., Van Neer, W., Sergant, Crombe,´ Ph., Sergant, J., Robinson, E. & De Reu, J., 2011. Hunter-gatherer J. & Crombe,´ Ph., 2013. Wood charcoal and seeds as indicators for animal hus- responses to environmental change during the Pleistocene–Holocene tran- bandry in a wetland site during the late Mesolithic/early Neolithic transition sition in the southern North Sea basin: Final Palaeolithic–Final Mesolithic period (Swifterbant culture, ca. 4600–4000 BC) in NW-Belgium. Vegetation land use in northwest Belgium. Journal of Anthropological Archaeology 30: History and Archaeobotany 22: 51–60. 454–471. Deforce, K., Bastiaens, J. & Crombe,´ Ph., 2014a. A reconstruction of middle Crombe,´ Ph., Van Strydonck, M., Boudin, M., Van den Brande, T., Derese, Holocene alluvial hardwood forests (Lower Scheldt River, N-Belgium) and their C., Vandenberghe, D., Van den Haute, P., Court-Picon, M., Verniers, J., exploitation during the Mesolithic-Neolithic transition period (Swifterbant Gelorini, V., Bos, J., Verbruggen, F., Antrop, M., Bats, M., Bourgeois, J., Culture, c. 4500–4000 BC). Quaternaire 25(1): 9–21. De Reu, J., De Maeyer, Ph., De Smedt, Ph., Finke, P., Van Meirvenne, M. Deforce, K., Storme, A., Bastiaens, J., Debruyne, S., Denys, L., Ervynck, A., & Zwertvaegher, A., 2012. Absolute dating (14C and OSL) of the formation Meylemans, E., Stieperaere, H., Van Neer, W. & Crombe,´ Ph., 2014b. Middle- of coversand ridges occupied by prehistoric hunter-gatherers in NW Belgium. Holocene alluvial forests and associated fluvial environments: a multi-proxy Radiocarbon 54(3–4): 715–726. reconstruction from the lower Scheldt, N Belgium. The Holocene 24(11): 1550– Crombe,´ Ph., De Smedt, Ph., Davies, N.S., Gelorini, V., Zwertvaegher, A., 1564. Langohr, R., Van Damme, D., Demiddele, H., Van Strydonck, M., Antrop, De Kraker, A.M.J., 2006. Flood events in the southwestern Netherlands and M., Bourgeois, J., De Maeyer, Ph., De Reu, J., Finke, P., Van Meirvenne, coastal Belgium, 1400–1953. Hydrological Sciences Journal–Journal des Sci- M. & Verniers, J., 2013. Hunter-gatherer responses to the changing environ- ences Hydrologiques 51(5): 913–929. ment of the Moervaart palaeolake (NW Belgium) during the Late Glacial and De Kraker, A. M. J., 2007. De ontwikkeling van het landschap. In: Wilssens, M. Early Holocene. Quaternary International 308–309: 162–177. (ed.): Singelberg: hetkasteel en het land van Beveren. Lannoo (Tielt): 14–51. Crombe,´ Ph., Verhegge, J., Deforce, K., Meylemans, E. & Robinson, E., De Langen, G. J., 1992. Middeleeuws Friesland: de economische ontwikkeling 2015. Wetland landscape dynamics, Swifterbant land use systems, and the van het gewest Oostergo in de vroege en volle Middeleeuwen. Noordhoff Mesolithic–Neolithic transition in the southern North Sea basin. Quaternary (Groningen): 395 pp. International 378: 119–133. Demey, D., Vanhoutte, S., Pieters, M., Deforce, K., Denys, L., Ervynck, A., Cryns, J., Noens, G., Allemeersch, L., Bats, M., Cruz, F., Jongepier, I., Laloo, Lentacker, A., Storme, A. & Van Neer, W., 2013. Een dijk en een woonplat- P., Rozek, J., Sergant, J., Soens, T., Verhegge, J. & Windey, S., 2014. form uit de Romeinse periode in Stene (Oostende). Relicta 10: 7–70. Verrebroek – Logistiek Park Waasland fase West. Een paleolandschappelijke De Moor, G., 2002. Bijdrage tot de Quairtairgeologische kartering in Vlaanderen. en archeologische prospectie d.m.v. boringen en proefsleuven (03/2013 – Geologica Belgica 5 (1–2): 37–50. 01/2014). GATE-rapport 73 (Bredene): 247 pp. De Moor, G. & van de Velde, D., 1995. Toelichting bij de Quartairgeologische D’Alpaos, A., Lanzoni, S., Mudd, S., Fagherazzi, S., D’Alpaos, A., Lanzoni, S., kaart Kaartblad (14) Lokeren Schaal 1:50 000. Vlaamse overheid, Departement Mudd, S. & Fagherazzi, S., 2006. Modeling the influence of hydroperiod Leefmilieu, Natuur en Energie, Dienst Natuurlijke Rijkdommen: 123 pp. and vegetation on the cross-sectional formation of tidal channels. Estuarine, De Muynck, M., 1976. Het bedolven kultuurlandschap in de polders: een Coastal and Shelf Science 69(3–4): 311–324. rekonstruktie uitgaande van luchtfoto’s, toegepast op de Oost-Vlaamse de Bont, C., 2014. Amsterdamse boeren. Een historische geografie van het gebied en enkele Zeeuws-Vlaamse Polders. PhD Thesis. Ghent University (Ghent): tussen de duinen en het gooi in de Middeleeuwen. Verloren (Hilversum): 266 pp. 352 pp. Denys, L. 1993. Paleoecologisch diatomeëenonderzoek van de holocene afzettin- De Brouwer, J., Crosato, A., Dankers, N., van Duin, W., Herman, P.M.J., van gen in de westelijke Belgische kustvlakte. PhD Thesis. University of Antwerp Raaphorst, W., Stive, M.J.F., Talmon, A.M., Verbeek, H., de Vries, M.B., (Antwerp): 479 pp. van der Wegen, M. & Winterwerp, J.C., 2001. Eco-morphodynamic processes Denys, L. & Baeteman, C., 1995. Holocene evolution of relative sea level and in the Rhine-Meuse-Scheldt delta and the Dutch Wadden Sea. Delft Hydraulics local mean high water spring tides in Belgium – a first assessment. Marine report Z2817: 93 pp. Geology 124: 1–19. De Clercq, W., 2009. Extrema Gallarium. Zeeland en Noorwest-Vlaanderen in het De Reu, J., Deweirdt, E., Crombe,´ Ph., Bats, M., Antrop, M., De Maeyer, P., De Imperium Romanum. Zeeuws Tijdschrift 58(3–4): 6–34. Smedt, Ph., Finke, P., Van Meirvenne, M., Verniers, J., Zwertvaegher, A. De Clercq, W. & Van Dierendonck, R., 2009. Extrema Galliarum: Noordwest- & Bourgeois, J., 2011. Les tombelles de l’âge du bronze en Flandre sablon- Vlaanderen en Zeeland in het Imperium Romanum. Verbond voor oudheid- neuse (nord-ouest de la Belgique): un status quaestionis. Archäologisches kundig bodemonderzoek in Oost-Vlaanderen (VOBOV-info): 34–75. Korrespondenzblatt 41(4): 491–505.

Dijkema, K.S., Beeftink, W.G., Doody, J.P., Gehu, J.M., Heydemann, B. & Rivas Kasse, C. K., 2002. Sandy aeolian deposits and environments and their relation Martinez, M., 1984. Salt marshes in Europe. Council of Europe, Nature and to climate during the Last Glacial Maximum and Lateglacial in northwest and environment series (Strasbourg) 30: 177 pp. central Europe. Progress in Physical Geography 26(4): 507–532. Dijkstra, J. & Zuidhoff, F.S., 2011. Kansen op de kwelder. Archeologisch onder- Kiden, P., 1989. Holocene water level movements in the lower Scheldt perima- zoek op en rond negen vindplaatsen in het nieuwe tracé van de Rijksweg 57 en rine area. In: Baeteman, C. (ed.): Quaternary sea-level investigations from de nieuwe rondweg ter hoogte van Serooskerke (Walcheren). ADC Monografie Belgium: a contribution to IGCP Project 200. Professional Paper, Geological (Amersfoort) 10: 35–58. Survey of Belgium (Brussels): 1–19. Gelorini, V., Meersschaert, L. & Van Roeyen, J.-P., 2003. Archeobotanisch on- Kiden, P., 1991. The Lateglacial and Holocene Evolution of the Middle and Lower derzoek van enkele laat- en postmiddeleeuwse archeologische contexten uit Scheldt, Belgium. In: Starkel, K.J. and Thornes, J.B. (eds): Temperate palaeo- de onderzoekszone Verrebroekdok (Beveren, Prov. Oost-Vlaanderen). Arche- hydrology. John Wiley & Sons (Chichester): 283–299. ologie in Vlaanderen VII 1999-2000: 200–224. Kiden, P., 1995. Holocene relative sea-level change and crustal movement in the Gelorini, V., Verleyen, E., Verbruggen, C. & Meersschaert, L., 2006. Paleo- southwestern Netherlands. Marine Geology 124: 21–41. ecologisch onderzoek van een Holocene sequentie uit het Deurganckdok te Kiden, P., 2006. De evolutie van de Beneden-Schelde in België en Zuidwest- Doel (Wase Scheldepolders, Noord-België). BELGEO 3: 243–264. Nederland na de laatste ijstijd. BELGEO 3: 279–294. Gottschalk, M.K.E., 1984. De Vier Ambachten en het Land van Saaftinge in de Kiden, P. & Verbruggen, C., 2001. Het verhaal van een rivier: de evolutie van middeleeuwen: een historischgeografisch onderzoek betreffende Oost Zeeuws- de Schelde na de laatste ijstijd. In: Bourgeois, J., Crombé, P., De Mulder, G. Vlaanderen c.a.. Van Gorcum (Assen, the Netherlands): 589 pp. & Rogge, M. (eds): Een in duik in het verleden. Schelde, Maas en Rijn in de Guns, P., 1975. Historische evolutie van het polderlandschap langs de linker pre- en protohistorie. Publicaties van het Provinciaal Museum van Zuid-Oost- Scheldeoever. Waterbouwkundig Laboratorium 1933-2008. Vlaamse Overheid, Vlaanderen – site Velzeke 4: 11–35. Departement Mobiliteit en Openbare werken: 78 pp. Knol, E., 2013. Moorkolonisation und Deichbau als Ursache von Flutkatastrophen: Heyse, I. & De Moor, G., 1979. Morphology of Würm Lateglacial and Holocene das Beispiel der nördlichen Niederlande. Siedlungs- und Küstenforschung im deposits in the Flemish valley (North Belgium). Acta Universitatis Ouluensis Südlichen Nordseegebiet 36: 157–170. A Scientiae Rerum Naturalium 82: 121–131. Leenders, K., 1986. 2000 jaar kustontwikkeling van Cap Griz Nez tot Hoek van Hol- Hijma, M.P. & Cohen, K.M., 2011. Holocene transgression of the Rhine land. Report nr. NZ-N-86.19. Rijkswaterstaat, taakgroep 1000 van het project river mouth area, The Netherlands/Southern North Sea: palaeogeogra- kustgenese (Rijswijk): 44pp. phy and sequence stratigraphy. Netherlands Journal of Geosciences 58: Leenders, K., 1989. Verdwenen venen: een onderzoek naar de ligging en ex- 1453–1485. ploitatie van thans verdwenen venen in het gebied tussen Antwerpen, Turn- Jacobs, P., De Ceukelaire, M., De Breuck, W. & De Moor, G., 1993. Geologis- hout, Geertruidenberg en Willemstad (1250–1570). Reeks Landschapsstudies, che kaart van België, Vlaams gewest, Kaartblad (14) Lokeren Schaal 1:50 Gemeentekrediet (Brussel): 351 pp. 000. Vlaamse overheid, Departement Leefmilieu, Natuur en Energie, Dienst Ligtendag, W.A., 1995. De Wolden en het water: de landschaps- en waterstaat- Natuurlijke Rijkdommen. sontwikkeling in het lage land ten oosten van de stad Groningen vanaf de Jacobs, P., Polfliet, T., De Ceukelaire, M. & Moerkerke, G., 2010. Geologische volle middeleeuwen tot c. 1870. Regio- en landschapsstudies van de Stichting kaart van België, Vlaams gewest, Kaartblad (1-7) Essen - Kapellen Schaal 1:50 historisch onderzoek en beleid II (Groningen): 368 pp. 000. Vlaamse overheid, Departement Leefmilieu, Natuur en Energie, Dienst Lunne, T., Robertson, P.K. & Powell, J.J.M., 1997. Cone penetration testing in Natuurlijke Rijkdommen. geotechnical practice. Spon Press Taylor & Francis Group (London): 305 pp. Janssens, W. & Ferguson, D.K., 1985. The palaeoecology of the Holocene sed- Mauri, A., Davis, B.A.S., Collins, P.M. & Kaplan, J.O., 2015. The climate of iments at Kallo, Northern Belgium. Review of Palaeobotany and Palynology Europe during the Holocene: a gridded pollen-based reconstruction and its 46: 81–95. multi-proxy evaluation. Quaternary Science Reviews 112: 109–127. Jongepier, I., Soens, T., Thoen, E., Eetvelde, V., Crombe,´ Ph. & Bats, M., 2011. Mayle, F. & Iriarte, J., 2014. Integrated palaeoecology and archaeology: a power- The brown gold: a reappraisal of medieval peat marshes in Northern Flanders ful approach for understanding Pre-Columbian Amazonia. Journal of Archae- (Belgium). Water History 3: 73–93. ological Sciences 51: 54–64. Jongepier, I., Soens, T. & Temmerman, S., 2012. Poldercartografie. De rol van Meersschaert, L., Van Roeyen, J.-P. & Verbruggen, C., 2006. Geomorfologisch, kaarten bij bedijkingen en landschapstransformatie. In: Ooghe, B., Goossens, geoarcheologisch, paleoecologisch en paleobotanisch onderzoek van de have- C. & Segers, Y. (eds): Van brouck tot dyckagie. Vijf eeuwen Wase polders. nuitbreidingswerken op de linker Scheldeoever ten noorden van Antwerpen. Abimo (Sint-Niklaas): 45–68. BELGEO 3: 183–203. Jongepier, I., Soens, T., Temmerman, S., Missiaen, T. & De Wit, B., 2015a. Meire, P. & Kuijken, E., 1988. Het land van Saeftinge, slikken en schorren: Assessing the planimetric accuracy of historical maps (sixteenth to nineteenth ecologische betekenis van getijdengebieden langs de Schelde. Water 43: 214– centuries). New methods and potential for coastal landscape reconstruction. 222. Cartographic Journal, in press (DOI: 10.1179/1743277414Y. 0000000095). Mercuri, A.M., Allevato, E., Arobba, D., Bandini Mazzanti, M., Bosi, G., Jongepier, I., Wang, C., Missiaen, T., Soens, T. & Temmerman, S., 2015b. Inter- Caramiello, R., Castiglioni, E., Carra, M.E., Celant, A., Costantini, L., Di tidal landscape response to dike breaching and (gradual) re-embankment: a Pasquale, G., Fiorentino, G., Florenzano, A., Guido, M., Marchesini, M., combined and historical and geomorphological reconstruction. Geomorphol- Mariotti Lippi, M., Marvelli, S., Miola, A., Montanari, C., Nisbet, R., Pena-˜ ogy 236: 64–78. Chocarro, L., Perego, R., Ravazzi, C., Rottoli, M., Sadori, L., Ucchesu, M.

& Rinaldi, R., 2014. Pollen and macroremains from Holocene archaeological Sergant, J., Crombe,´ Ph. & Perdaen, Y., 2006. The ‘invisible’ hearths: a contribu- sites: a dataset for the understanding of the bio-cultural diversity of the tion to the discernment of Mesolithic non-structured surface hearths. Journal Italian landscape. Review of Palaeobotany and Palynology 218: 250–266. of Archaeological Science 33: 999–1007. Meylemans, E., Bogemans, F., Storme, A., Perdaen, Y., Verdurmen, I. & Sergant, J., Bats, M., Noens, G., Lombaert, L. & D’Hollander, D., 2007. Voorlop- Deforce, K., 2013. Lateglacial and Holocene fluvial dynamics in the Lower ige resultaten van noodopgravingen in het afgedekte dekzandlandschap van Scheldt basin (N-Belgium) and their impact on the presence, detection and Verrebroek – Aven Ackers (Mesolithicum, Neolithicum). Notae Praehistoricae preservation potential of the archaeological record. Quaternary International 27: 101–107. 308–309: 148–161. Sier, M.M. (ed.), 2003. Ellewoutsdijk in de Romeinse tijd. ADC rapport 200 (Amers- Mijs, M., 1973. De landschapsgeschiedenis van de Scheldepolders ten noorden foort): 198 pp. van Antwerpen. Bijdrage tot de historische geografie van de Scheldepold- Snacken, F., 1964. De ontwikkeling van het Scheldepolderlandschap. Verslagboek ers. Tijdschrift van de Belgische Vereniging voor Aardrijkskundige Studies 4e Int. Havencongres Antwerpen 22–27 juni 1964: 485–490. (D/1973/0468/2): 40–124. Soens, T., 2013. The origins of the Western Scheldt. Environmental transforma- Minnaert, G. & Verbruggen, C., 1986. Palynologisch onderzoek van een veen- tion, storm surges and human agency in the Flemish coastal plain (1250– profiel uit het Doeldok te Doel. Bijdragen van de Archeologische Dienst 1600). In: Thoen, E., Borger, G. J., De Kraker, A., Soens, T., Tys, D., Vervaet, Waasland 1: 201–208. L. & Weerts, H. (eds): Landscapes or seascapes? The history of the coastal Missiaen, T., Slob, E. & Donselaar, M.E., 2008. Comparing different shallow environment in the North Sea area reconsidered. Comparative Rural History geophysical methods in a tidal estuary, Verdronken Land van Saeftinge, of the North Sea Area 13: 287–312. Western Scheldt, The Netherlands. Netherlands Journal of Geosciences 87(2): Soens, T. & Thoen, E., 2009. Mais où sont les tourbières d’antan? Géographie, 151–164. chronologie et stratégieseconomiques ´ du tourbage en Flandre Maritime (12e– Missiaen, T., Verhegge, J., Heirman, K. & Crombe,´ Ph., 2015. Potential of Cone 16e siècles). Aestuaria, Histoire et Terres Humides 14: 45–60. Penetrating Testing for mapping deeply buried palaeolandscapes in the con- Soens, T., Tys, D. & Thoen, E., 2014. Landscape transformation and social change text of archaeological surveys in polder areas. Journal of Archaeological Sci- in the North Sea Polders, the example of Flanders (1000–1800 AD). Siedlungs- ence 55: 174–187. forschung: Archäologie, Geschichte, Geographie 31: 133–160. Moore, P.D., Webb, J.A. & Collinson, M.E., 1991. Pollen analysis (2nd edition). Thoen, H. (ed.), 1989. Temse en de Schelde van ijstijd tot de Romeinen. Gemeen- Cambridge University Press (Cambridge): 216 pp. tekrediet (Brussel): 128 pp. Munaut, A.-V., 1967. Recherches paléo-écologiques en Basse et Moyenne Bel- Van Dam, P., 2001. Sinking peat bogs: environmental change in Holland, 1350– gique. Acta Geographica Lovaniensia 6: 191 pp. 1550. Environmental History 6: 32–46. Nelle, O., Dreibrodt, S. & Dannath, Y., 2010. Combining pollen and charcoal: Vandenbruwaene, W., Meire, P. & Temmerman, S., 2012. Formation and evo- evaluating Holocene vegetation composition and dynamics. Journal of Ar- lution of a tidal channel network within a constructed tidal marsh. Geomor- chaeological Science 37: 2126–2135. phology 151/152: 114–125. Ovaa, I., Van der Meer, K. & Steur, G.G.L., 1957. De bodemgesteldheid van Van der Spek, A., 1994. Large-scale evolution of Holocene tidal basins in the Westelijk Zeeuws-Vlaanderen. Stichting voor Bodemkartering (Wageningen), Netherlands. PhD Thesis. Utrecht University (Utrecht): 191 pp. rapport nr. 455. Van der Spek, A. & Beets, D., 1992. Mid-Holocene evolution of a tidal basin Perdaen, Y., Sergant, J. & Crombe,´ Ph., 2004. Early Mesolithic landscape-use in the western Netherlands: a model for future changes under conditions of and site-use in northwestern Belgium: the evidence from Verrebroek “Dok”. accelerated sea-level rise? Sedimentary Geology 80: 185–197. In:Crombé, Ph. (ed.): Proceedings of the symposium “Landscape-Use during Van Gerven, R., 1977. De Scheldepolders van de Linkeroever (Land van Waas en the Final Palaeolithic and Mesolithic in NW-Europe; The Formation of Exten- Land van Beveren): bijdrage tot de geschiedenis van natuur-land-volk. Sint sive Sites and Site-Complexes”. British Archaeological Reports, International Niklaas: 780 pp. Series 1302: 11–18. Van Neer, W., Ervynck, A., Lentacker, A., Bastiaens, J., Deforce, K., Sergant, Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Ramsey, C.B., J. & Crombe,´ Ph., 2013. Hunting, gathering, fishing and herding: animal Buck, C.E., Cheng, H., Edwards, R.L., Friedrich, M., Grootes, P., Guilderson, exploitation in Sandy Flanders (NW Belgium) during the second half of the T., Haflidason, H., Hajdas, H., Hatte,´ C., Heaton, T., Hoffmann, D., Hogg, 5th millennium BC. Journal of Environmental Archaeology 18(2): 87–101. A., Hughen, K., Kaiser, K., Kromer, B., Manning, S., Niu, M., Reimer, R., Van Roeyen, J.-P. (ed.), 2007. Paleolandschappelijk en archeologisch onderzoek Richards, D., Scott, E., Southon, J., Staff, R., Turney, C. & van der Plicht, van de te realiseren natuurcompensatiezone “Weidevogelgebied Doelpolder J., 2013. IntCal13 and Marine13 radiocarbon age calibration curves 0–50,000 Noord en Kreek in het kader van de bouw van het Containergetijdendok-West years cal BP. Radiocarbon 55(4): 1869–1887. (Gemeente Beveren). Eindrapport: paleolandschappelijke en archeologische Rippon, S., 2000. The transformation of coastal wetlands: exploitation and man- screening aan de hand van boringen en inventarisatie (parentheses 1 en 2). agement of marshland landscapes in north west Europe during the Roman Archeologische Dienst Waasland (Sint-Niklaas): 102 pp. and medieval periods. Oxford University Press (Oxford): 332 pp. Van Rummelen, F., 1965. Toelichtingen bij de Geologische Kaart van Nederland, Rippon, S., 2002. Romney Marsh: evolution of the historic landscape and its 1:50.000; Bladen Zeeuwsch-Vlaanderen Wesxt en Oost. Rijks Geologische Di- wider significance. In: Long, A., Hipkin, S. & Clarke, H. (eds): Romney Marsh: enst (Haarlem): 79 pp. coastal and landscape change through the ages. Oxford University School of Van Strydonck, M., 2005. Radiocarbon dating. In: Crombé, Ph. (ed.): The last Archaeology Monograph (Oxford) 56: 84–100. hunter-gatherer-fishermen in sandy Flanders (NW Belgium). The Verrebroek

and Doel excavation projects (Vol. 1: Palaeo-environment, chronology and Vos, P.C. & Knol, E., 2015. Holocene landscape reconstruction of the Wadden features). Archaeological Reports Ghent University 3, Academia Press (Ghent): Sea area between Marsdiep and Weser. In: Vos, P., Origin of the Dutch coastal 180–212. landscape. Long-term evolution of the Netherlands during the Holocene, de- Verbruggen, C. & Denys, L., 1995. Early tidal influence on the Lower Schelde, scribed and visualised in national, regional and local palaeographical maps Belgium. Aardkundige Mededelingen 6: 167–169. series. Barkhuis (Groningen): 202–229. Verbruggen, C., Denys, L. & Kiden, P., 1991. Paleo-ecologische en geomorfol- Vos, P.C. & van Heeringen, R.M., 1997. Holocene geology and occupation history ogische evolutie van Laag- en Midden-België tijdens het Laat-Kwartair. De of the province of Zeeland. In: Fischer, M.M. (ed.): Holocene evolution of Zee- Aardrijkskunde 3: 357–376. land (SW Netherlands). Mededelingen Nederlands Instituut voor Toegepaste Verbruggen, C., Denys, L. & Kiden, P., 1996. Palaeoecological events in Belgium Geowetenschappen 59: 5–109. during the last 13,000 years with special reference to sandy Flanders. In: Vos, P.C., Moree, J. & Zeiler, F.D., 2002. Delta-2003, 5000 jaar terugblik, kaartat- Berglund, B.E., Ralska-Jasiewiczowa, M. & Wright, H.E. (eds.): Palaeoecologi- las met toelichting. Landschapsreconstructie van de kustdelta van ZW Ned- cal events during the last 15000 years: regional syntheses of palaeoecological erland in opdracht van het project GEOMOD van het RIKZ van het Ministerie studies of lakes and mires in Europe. John Wiley & Sons (Chichester): 553–574. van Verkeer en Waterstaat. TNO rapport NITG 02-096-B. Verhegge, J., 2015. Spatial and chronological prehistoric landscape reconstruc- Vos, P.C., de Koning, R. & van Eerden, R., 2015. Landscape history of the Oer-IJ tion using geoarchaeological methods in the lower Scheldt floodplain (NW tijdal system, Noord-Holland (The Netherlands). In: Vos, P. (ed.), Origin of Belgium). PhD Thesis. Ghent University (Ghent): 247 pp. the Dutch coastal landscape. Long-term evolution of the Netherlands dur- Verhegge, J., Missiaen, T., Van Strydonck, M. & Crombe,´ Ph., 2014. Chronol- ing the Holocene, described and visualised in national, regional and local ogy of wetland hydrological dynamics and the Mesolithic-Neolithic transition palaeographical maps series. Barkhuis (Groningen): 98–201. along the Lower Scheldt: a Bayesian approach. Radiocarbon 56: 883–898. Wang, C. & Temmerman, S., 2013. Does bio-geomorphic feedback lead to abrupt Verhoeve, A. & Verbruggen, C., 2006. Het Meetjesland. Bodem en landschap in shifts between alternative landscape states? An empirical study on intertidal historisch perspectief. BELGEO 7: 205–218. flats and marshes. Journal of Geophysical Research – Earth Surface 118: 229– Vos, P.C., 2015. Origin of the Dutch coastal landscape. Long-term evolution of the 240. Netherlands during the Holoce, described and visualised in national, regional Wilssens, M., Bartholomieux, B., De Kraker, A., De Meulemeester, J., Poschet, and local palaeographical maps series. Barkhuis (Groningen): 359 pp. K., Van Daele, R., Verelst, D. & Willems, R., 2007. Singelberg, Het kasteel Vos, P.C. & de Wolf, H., 1993. Diatoms as a tool for reconstructing sedimentary en het land van Beveren. Lannoo (Tielt): 270 pp. environments in coastal wetlands; methodological aspects. Hydrobiologia 269: Woodruff, J. D., Irish, J. L. & Camargo, S. J., 2013. Coastal flooding by tropical 285–296. cyclones and sea-level rise. Nature 504: 44–52.